A Good Practices Guide for Digital Image Correlation

A Good Practices Guide for

Digital Image Correlation

Standardization, Good Practices, and Uncertainty Quantication Committee

October, 2018

Safety Disclaimer

This guide does not address the health or safety concerns regarding applying DIC

in a mechanical testing or laboratory environment. It is the responsibility of the

laboratory and user to determine the appropriate safety and health requirements.

Some rights reserved. This publication may be reproduced, distributed, or transmitted

in any form or by any means, including photocopying, recording, or other electronic or

mechanical methods, but only without alteration and with full attribution to the Inter-

national Digital Image Correlation Society (iDICs). Exception is given in the case of

brief quotations embodied in other documents with full attribution to the International

Digital Image Correlation Society, but not in a way that suggests endorsement of the

other document by the International Digital Image Correlation Society. For permission

requests, contact the International Digital Image Correlation Society at [email protected].

DOI: 10.32720/idics/gpg.ed1/print.format

Electronic copies of this guide are available at www.idics.org.

Suggested citation: International Digital Image Correlation Society, Jones, E.M.C. and

Iadicola, M.A. (Eds.) (2018). A Good Practices Guide for Digital Image Correlation.

https://doi.org/10.32720/idics/gpg.ed1/print.format.

iii

About this Guide

The International Digital Image Correlation Society (iDICs) was founded in 2015 as a

nonproﬁt scientiﬁc and educational organization committed to training and educating

users of digital image correlation (DIC) systems. iDICs is composed of members from

academia, government, and industry, and develops world-recognized DIC training and

certiﬁcations to improve industry practice of DIC for general applications, with em-

phasis on both research and establishing standards for DIC measurement techniques.

More information can be found at www.idics.org.

To support this mission, the iDICs Standardization, Good Practices, and Uncer-

tainty Quantiﬁcation Committee was formed in part to develop guidelines for DIC

practitioners. Details of the entire development and review process can be obtained

through iDICs ([email protected]), but they are summarized here. The working group on

Good Practices, Reporting Requirements and Terminology (a subset of the committee)

developed this Good Practices Guide for DIC. The working group was composed of

expert DIC practitioners (see below), including representatives from many commercial

DIC software packages, with diverse experience using DIC in a myriad of applications.

After a ﬁnal draft of the guide was completed by the working group, a public

comment period was opened in November 2017 through January 2018, during which

any DIC practitioner could opt-in to review the Guide. In total, 100 people opted-in

to the review process, 56 of whom returned oﬃcial votes. Of the 56 received votes, 23

people voted “Approve without comment”, 32 people voted “Approve with comments

and suggested revisions”, and 1 person voted “Disapprove with comments (at least

one technical) and suggested revisions”. Over 500 comments were received (over 130

of which were technical comments), and the working group addressed each, either

through revising the Guide, or through a written rebuttal. After that revision, the

ﬁnal version of the Guide and the working group responses to the comments were

reviewed and approved by some of the members of the iDICs Executive Board, who

did not participate in either the working group or the public comment period.

Editor and Chair of Working Group

Elizabeth M. C. Jones, Ph.D., Senior Member of Technical Staﬀ, Sandia National

Laboratories, United States of America, Email: [email protected]

Co-Editor

Mark A. Iadicola, Ph.D., Staﬀ Scientist, National Institute of Standards and Tech-

nology, United States of America

Working Group Members

Rory P. Bigger, Senior Research Engineer, Southwest Research Institute, United

States of America

Benoˆıt Blaysat, Ph.D., Assistant Professor, Universit´e Clermont Auvergne, France

Christofer Boo, Product Manager, Image Systems, Sweden

Manuel Grewer, Ph.D., Product Manager, LaVision GmbH, Germany

Jun Hu, Ph.D., Engineer, AK Steel, United States of America

Amanda R. Jones, Ph.D., Senior Member of Technical Staﬀ, Sandia National Lab-

oratories, United States of America

Markus Klein, GOM GmbH, Germany

Pascal Lava, Ph.D., Managing Director, Match ID, Belgium

Mark Pankow, Ph.D., Assistant Professor, North Carolina State University, United

States of America

Kavesary Raghavan, Ph.D., PE, Senior Staﬀ Engineer, AK Steel Corporation,

United States of America

Phillip L. Reu, Ph.D., Principal Member of Technical Staﬀ, Sandia National Lab-

oratories, United States of America

Timothy Schmidt, Trilion/GOM, United States of America

Thorsten Siebert, Ph.D., R&D Manager, Dantec Dynamics GmbH, Germany

Micah Simonsen, Correlated Solutions, United States of America

Andrew Trim, Materials Engineering and Structural Dynamicist, Atomic Weapons

Establishment, United Kingdom

Daniel Z. Turner, Ph.D., Center for Computing Research, Sandia National Labo-

ratories, United States of America

Alessandro F. Vieira, Senior Instrumentation Engineer, Boeing, United States of

America

Thorsten Weikert, GOM GmbH, Germany

Conventions

In this guide, certain items are highlighted, set aside, or labeled separately from the

main body based on the following conventions.

Recommendation

“Recommendations” are suggestions about speciﬁc actions a DIC practitioner should

take, or speciﬁc decisions a DIC practitioner should make. These suggestions are

based on the collective experience and expertise of the working group members. Rec-

ommendations are intended to be ideal suggestions; in other words, for ideal DIC

measurements within the scope of this document, a DIC practitioner would follow all

recommendations.

Tip

“Tips” provide supplementary information that can be useful to a DIC practitioner,

helping to design and execute DIC measurements. “Tips” are typically background

information, targeted to either inexperienced DIC practitioners, or to experienced DIC

practitioners working with advanced setups. “Tips” are diﬀerentiated from “Recom-

mendations” in that “Tips” do not imply a speciﬁc action or decision that a DIC

practitioner should follow.

Caution

“Cautions” provide information about events, decisions, or features that could have

negative impacts on DIC measurements. “Cautions” are often followed by “Recom-

mendations”, which provide information on avoiding or mitigating the negative event,

decision, or feature.

Footnote

Footnotes are reserved for supplementary information that is outside of the scope of

this edition of the guide. Their primary purpose is to inform the reader when the

guidelines given in this guide are not applicable, to ensure that the guidelines are

applied appropriately. The remarks made in footnotes are brief, because the pieces of

information contained in the footnotes are outside of the scope of this edition of the

guide.

Appendix

Appendices are reserved for supplementary information that, due to length or com-

plexity, would clutter and defocus the main body of text.

vii

viii

Contents

Front Matter ii

Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii

About this Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

List of Tips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

List of Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

List of Cautions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv

1 Introduction 1

1.1 Aims and Basic Principles of DIC . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Scope of this Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Scope for Common Mechanical Tests with DIC Measurements . . . . . . 3

2 Design of DIC Measurements 5

2.1 Measurement Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.1 Quantity-of-Interest . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.2 Region-of-Interest . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.3 Field-of-View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.4 Position Envelope for Hardware . . . . . . . . . . . . . . . . . . . 6

2.1.5 2D-DIC vs Stereo-DIC . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1.6 Stereo-Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1.7 Depth-of-Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1.8 Spatial Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1.9 Noise-Floor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1.10 Frame Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1.11 Exposure Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.1.12 Synchronization and Triggering . . . . . . . . . . . . . . . . . . . 10

2.2 Equipment and Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.2.1 Camera and Lens Selection . . . . . . . . . . . . . . . . . . . . . 11

2.2.2 Camera and Lens Mounting . . . . . . . . . . . . . . . . . . . . . 13

2.2.2.1 General Characteristics of Mounting System . . . . . . 13

2.2.2.2 Types of Mounting Systems . . . . . . . . . . . . . . . . 17

2.2.3 Aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.2.4 Lighting and Exposure . . . . . . . . . . . . . . . . . . . . . . . . 18

2.2.4.1 Type of Lights . . . . . . . . . . . . . . . . . . . . . . . 19

2.2.4.2 Light Mounting . . . . . . . . . . . . . . . . . . . . . . 19

2.2.4.3 Contrast, Intensity, and Gain . . . . . . . . . . . . . . . 20

2.2.5 Hardware Heating . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.3 DIC Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.3.1 Type of DIC Patterns . . . . . . . . . . . . . . . . . . . . . . . . 22

2.3.2 General Characteristics of DIC Patterns . . . . . . . . . . . . . . 23

2.3.2.1 Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.3.2.2 Variation . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.2.3 Density . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.2.4 Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.3.2.5 Reﬂections . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.3.3 Characteristics of Applied Patterns . . . . . . . . . . . . . . . . . 26

2.3.3.1 Compliance . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.3.2 Bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.3.3 Fidelity . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

2.3.3.4 Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.3.4 Patterning Techniques . . . . . . . . . . . . . . . . . . . . . . . . 28

3 Preparation for the Measurements 31

3.1 Pre-Calibration Routine . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.1.1 Review of Test Procedure . . . . . . . . . . . . . . . . . . . . . . 31

3.1.2 Cleanliness of Equipment . . . . . . . . . . . . . . . . . . . . . . 31

3.1.3 Camera Warm-Up . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.1.4 Synchronization . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.1.5 Application of the DIC Pattern . . . . . . . . . . . . . . . . . . . 34

3.1.6 Pre-Calibration Review of System . . . . . . . . . . . . . . . . . 34

3.1.6.1 Position Test Piece and Cameras . . . . . . . . . . . . . 34

3.1.6.2 Verify Optical System . . . . . . . . . . . . . . . . . . . 35

3.1.6.3 Lock Adjustable Components . . . . . . . . . . . . . . . 35

3.1.6.4 Review Images . . . . . . . . . . . . . . . . . . . . . . . 35

3.1.6.5 Accept the DIC System . . . . . . . . . . . . . . . . . . 36

3.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.2.1 Purpose of Calibration . . . . . . . . . . . . . . . . . . . . . . . . 36

3.2.2 General Calibration Steps . . . . . . . . . . . . . . . . . . . . . . 37

3.2.2.1 Select Calibration Target . . . . . . . . . . . . . . . . . 37

3.2.2.2 Clear Working Space . . . . . . . . . . . . . . . . . . . 38

3.2.2.3 Adjust Lighting . . . . . . . . . . . . . . . . . . . . . . 39

3.2.2.4 Acquire Calibration Images . . . . . . . . . . . . . . . . 40

3.2.2.5 Calibrate System . . . . . . . . . . . . . . . . . . . . . . 42

3.2.2.6 Review Calibration Results . . . . . . . . . . . . . . . . 42

3.2.2.7 Review Calibration Parameters . . . . . . . . . . . . . . 43

3.3 Post-Calibration Routine . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.3.1 Images for Calibration Veriﬁcation and Noise-Floor Analysis . . 45

3.3.1.1 Reset System . . . . . . . . . . . . . . . . . . . . . . . . 45

3.3.1.2 Adjust Lighting . . . . . . . . . . . . . . . . . . . . . . 45

3.3.1.3 Acquire Static Images . . . . . . . . . . . . . . . . . . . 45

3.3.1.4 Review Images . . . . . . . . . . . . . . . . . . . . . . . 45

3.3.1.5 Acquire Rigid-Body-Motion Images . . . . . . . . . . . 46

3.3.2 Veriﬁcation of Calibration . . . . . . . . . . . . . . . . . . . . . . 46

3.3.2.1 Intrinsic parameters . . . . . . . . . . . . . . . . . . . . 47

3.3.2.2 Extrinsic parameters . . . . . . . . . . . . . . . . . . . . 47

3.3.2.3 Absolute distances . . . . . . . . . . . . . . . . . . . . . 48

3.3.3 Post-Calibration Review of System . . . . . . . . . . . . . . . . . 49

3.3.3.1 Noise-Floor . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.3.3.2 Heat Waves . . . . . . . . . . . . . . . . . . . . . . . . . 49

3.3.3.3 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.3.3.4 Other Veriﬁcations . . . . . . . . . . . . . . . . . . . . . 50

4 Execution of the Test with DIC Measurements 51

5 Processing of DIC Images 53

5.1 DIC Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.2 User-Deﬁned Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.2.1 Reference Image . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.2.2 Pre-Filtering of Images . . . . . . . . . . . . . . . . . . . . . . . 54

5.2.3 Subset Shape Function . . . . . . . . . . . . . . . . . . . . . . . . 55

5.2.4 Interpolant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5.2.5 Subset Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.2.6 Step Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.2.7 Thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.3 Strain Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.3.1 Virtual Strain Gauge (VSG) . . . . . . . . . . . . . . . . . . . . 57

5.3.2 Examples of Strain Calculation Methods . . . . . . . . . . . . . . 57

5.3.2.1 Subset Shape Function . . . . . . . . . . . . . . . . . . 58

5.3.2.2 Finite-Element Shape Functions . . . . . . . . . . . . . 58

5.3.2.3 Strain Shape Function . . . . . . . . . . . . . . . . . . . 58

5.3.2.4 Spline Fit . . . . . . . . . . . . . . . . . . . . . . . . . . 59

5.4 Uncertainty Quantiﬁcation . . . . . . . . . . . . . . . . . . . . . . . . . . 59

5.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

5.4.2 Variance Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

5.4.3 Bias Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.4.4 Trade-Oﬀ Between Noise and Bias . . . . . . . . . . . . . . . . . 62

5.4.5 Virtual Strain Gauge Study . . . . . . . . . . . . . . . . . . . . . 62

6 Reporting Requirements 65

6.1 DIC Hardware Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.1.1 Required . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6.1.2 Recommended . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

6.2 DIC Analysis Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 66

6.2.1 Required . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

6.2.2 Recommended . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

7 Glossary and Acronyms 69

7.1 Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

7.2 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

References 76

A Checklist and Flow Chart for DIC Measurements and Analysis 81

B Focal Length, Field-of-View, and Stand-Oﬀ Distance 89

Notes 91

List of Figures

2.1 Recommended mounting orientations of cameras for stereo-DIC. . . . . 15

A.1 Flow chart of main steps for DIC measurements (part 1). . . . . . . . . 86

A.2 Flow chart of main steps for DIC measurements (part 2). . . . . . . . . 87

List of Tables

B.1 Example reference table for SOD, focal length, and FOV. . . . . . . . . 90

List of Tips

Tip 2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Tip 2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Tip 2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Tip 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Tip 2.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Tip 2.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Tip 2.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Tip 2.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Tip 2.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Tip 2.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Tip 2.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Tip 2.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Tip 2.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Tip 2.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Tip 2.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Tip 2.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Tip 2.17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Tip 2.18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Tip 2.19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

xii

Tip 2.20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Tip 2.21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Tip 2.22 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Tip 2.23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Tip 2.24 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Tip 2.25 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Tip 2.26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Tip 2.27 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Tip 2.28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Tip 2.29 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Tip 2.30 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Tip 2.31 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Tip 3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Tip 3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Tip 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Tip 3.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Tip 3.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Tip 3.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Tip 3.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Tip 3.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Tip 3.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Tip 3.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Tip 3.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Tip 3.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Tip 3.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Tip 3.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Tip 3.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Tip 5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Tip 5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Tip 5.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

Tip 5.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Tip 5.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Tip 6.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

List of Recommendations

Recommendation 2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Recommendation 2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Recommendation 2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Recommendation 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Recommendation 2.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Recommendation 2.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Recommendation 2.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Recommendation 2.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Recommendation 2.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Recommendation 2.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Recommendation 2.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

xiii

Recommendation 2.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Recommendation 2.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Recommendation 2.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Recommendation 2.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Recommendation 2.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Recommendation 2.17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Recommendation 2.18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Recommendation 2.19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Recommendation 2.20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Recommendation 3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Recommendation 3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Recommendation 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Recommendation 3.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Recommendation 3.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Recommendation 3.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Recommendation 3.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Recommendation 3.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Recommendation 3.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Recommendation 3.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Recommendation 3.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Recommendation 3.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Recommendation 3.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Recommendation 3.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Recommendation 3.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Recommendation 3.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Recommendation 3.17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Recommendation 3.18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Recommendation 5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Recommendation 5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

Recommendation 5.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Recommendation 6.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

List of Cautions

Caution 2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Caution 2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Caution 2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Caution 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Caution 2.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Caution 2.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Caution 2.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Caution 2.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Caution 2.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Caution 2.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Caution 2.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Caution 2.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Caution 2.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

xiv

Caution 2.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Caution 2.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Caution 2.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Caution 2.17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Caution 2.18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Caution 2.19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Caution 2.20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Caution 2.21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Caution 2.22 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Caution 2.23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Caution 2.24 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Caution 2.25 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Caution 3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Caution 3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Caution 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Caution 3.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Caution 3.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Caution 3.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Caution 3.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Caution 3.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Caution 3.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Caution 3.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Caution 3.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Caution 3.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Caution 3.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Caution 3.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Caution 3.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Caution 3.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Caution 3.17 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Caution 4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Caution 5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Caution 5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Caution 5.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

xvi

1 — Introduction

1.1 Aims and Basic Principles of DIC

Within the scope of this guide, DIC is an optically-based technique used to measure

the evolving full-ﬁeld 2D or 3D coordinates on the surface of a test piece throughout a

mechanical test. The measured coordinate ﬁelds can be used to calculate derived ﬁeld

quantities-of-interest (QOIs), such as displacements, strains, strain rates, velocities,

and curvatures. Because DIC is a non-contact technique that is independent of the

material being tested or the length-scale of interest, it can be used in a wide variety

of applications to investigate and characterize the deformation of solids. Some com-

mon materials that are tested include metals, polymers, concrete, geological samples,

biological tissues, battery electrodes, explosives, etc. and test pieces range from, for ex-

ample, small coupons used in tensile tests up to entire sub-assemblies of aircraft. This

versatility has led to a plethora of methodologies and software codes, both commercial

and independently developed, to utilize the data captured from a DIC measurement.

To begin, the basic concepts of DIC are introduced here. For a more complete

introduction to DIC, one should consult [42]. For a practical guide to getting started

with DIC, the reader is also directed to the series of articles in [16]. At the core level,

DIC estimates full-ﬁeld coordinates and displacements from a sequence of digital images

taken of a pattern on the surface of a test piece, by solving an optimization problem,

typically based on a transport model such as optical ﬂow. A fundamental assumption

in DIC measurements is that the pattern on the surface of the test piece, either natural

or applied, follows the deformation of the underlying test piece. Thus, the images of the

test piece taken throughout the test can be correlated to produce full-ﬁeld coordinates

representative of the shape, motion and deformation of the surface of the test piece.

2D coordinates of the surface can be measured using a single camera system, and

this is referred to as 2D-DIC. 3D coordinate measurements of the surface require a

minimum of two cameras

oriented at a stereo-angle to perform 3D photogrammetry

in addition to image correlation; this is called stereo-DIC.

Before measurements are

made, the camera/lens system is calibrated by imaging features of known separation

lengths (i.e. a calibration target). This calibration allows DIC software to correct

Stereo-DIC can be performed with a single camera using stereo optics, where the left and right

halves of the detector are taken as the left and right images. This is an advanced topic, though, and

beyond the scope of this edition of the guide.

Stereo-DIC is often commonly called 3D-DIC. However, to avoid confusion with volumetric-DIC,

this Guide recommends the term stereo-DIC instead of 3D-DIC. See the Glossary entry “Digital Image

Correlation” for more information.

for lens distortions and, for stereo-DIC, provides the location and orientation of the

cameras in space with respect to each other, and to the test piece.

There are many types of software developed to perform this correlation, but the

two most common categories are local and global DIC methods. In a local method,

the coordinate solution at a point depends only on a small subset of the image in the

vicinity of that point, but is otherwise independent of the solution at all other points

of interest. In a global method, the solution at one point has some dependence on the

solution at other points in the vicinity of the point of interest. For the most part, the

content of this guide is applicable to both methods, especially concerning the Design of

DIC Measurements (Ch. 2), Preparation for the Measurements (Ch. 3), and Execution

of the Test with DIC Measurements (Ch. 4); however, this guide is written from the

perspective of local DIC when discussing the Processing of DIC Images (Ch. 5).

In brief, a software code analyzes a user-deﬁned region-of-interest (ROI) within the

images, which contains a set of interrogation, or measurement, points. In local DIC,

each interrogation point is centered within a subset of the image. The interrogation

points are typically deﬁned at some regular spacing (step size), such that neighboring

subsets may (or may not) overlap. The subsets are numerically correlated from the

reference image (before motion/deformation) to each subsequent image (during mo-

tion/deformation). This correlation is performed by ﬁrst approximating the pattern in

each subset using an interpolant function, and then allowing that function to deform

from the reference image based on a subset shape function. A matching criterion in

conjunction with subset weights is used to match each subset in the reference image

with the corresponding subset in the deformed images. In stereo-DIC, the matching

criterion, along with the parameters of the stereo-system calibration, are used to match

subsets from one of the cameras to the other camera. The result of the correlation is

the measured coordinates of the center of each subset.

The calculation of derived ﬁeld quantities is the ﬁnal step in many DIC process-

ing schemes. The most commonly derived quantities from these coordinate ﬁelds are

probably strains, though DIC provides access to other QOIs, such as curvature, ve-

locity, and acceleration. The minimum resolution (also called the noise-ﬂoor) of the

QOIs, as well as potential bias errors, are tied to both the measurement setup (e.g.

camera selection, image contrast, DIC pattern feature size), and the data processing

parameters (e.g. subset size, subset shape function, virtual strain gauge). Therefore,

determination of the resolution of the QOIs through uncertainty quantiﬁcation analysis

completes the DIC data processing. This leaves the user with a full-ﬁeld description

of displacements, and/or derived quantities, of a test piece subjected to a mechanical

test, as well as the uncertainties of those measurements.

1.2 Scope of this Guide

The purpose of this document is to provide good-practice guidelines for conducting

DIC measurements in conjunction with mechanical testing of a planar test piece. This

guide is designed to be both a primer training document geared towards new practi-

tioners of DIC (supplementing vendor-based or other formal training and hardware-

and software-speciﬁc documentation), as well as a reference for experienced users, to

refresh their fundamentals knowledge and skill sets and assist them in troubleshooting

DIC measurements. Appendix A provides a checklist of the major points to consider

when designing, executing, and analyzing DIC measurements, while Fig. A.1 illustrates

the steps of a typical mechanical test with DIC measurements in graphical form. De-

tails for each step of the checklist and the ﬂow chart are presented in the body of each

section of this guide. The goal of this guide is to aid DIC practitioners in achieving

well planned, well executed, well analyzed, and well documented DIC measurements.

Note that this guide does not provide any guidelines for the mechanical test itself; it

focuses only on the complementary DIC measurements.

In developing this guide, we strove to include as many instructive and diagnostic

suggestions as possible, while still keeping the document general, and independent of

speciﬁc hardware or software packages. As this document is a guide and not a standard,

the guidelines presented here are not strict requirements for DIC measurements, but

rather are suggestions for good practices. However, some guidelines are considered to be

crucial for reliable and trustworthy measurements, while others are recommendations

that serve to increase conﬁdence in the measurements. This guide attempts to delineate

between crucial and recommended guidelines, and provide cautionary notes about the

consequences of omitting each guideline.

This guide focuses on good practices for DIC measurement setup, image correlation,

and basic post-processing of DIC data for strain computations. It does not cover other

data processing, such as velocity, acceleration, curvature, etc., nor speciﬁc data analysis

applications that utilize DIC data, such as Finite Element Model (FEM) validation,

material identiﬁcation, etc.

The scope of this edition of the good practices guide is limited to common laboratory

test conditions, as outlined in Sec. 1.3. Our intention is to incorporate good practices

for complex test conditions, and their associated additional challenges, in a future

edition of this guide.

1.3 Scope for Common Mechanical Tests

with DIC Measurements

This guide applies to the following conditions found in typical mechanical test arrange-

ments and associated DIC setups:

Test piece size of approximately 50 mm to 1 m

Planar test pieces undergoing nominally planar motion and/or deformation

Strain range of up to approximately 60% equivalent strain

General purpose laboratory testing with a well-controlled environment (e.g. room

temperature of 15–25

C, and minimal vibrations)

No special environmental conditions (e.g. no environmental chambers, no water

tanks or pressurized vessels, no windows or viewports, no explosions or shock

waves)

Optical-based images (no images based on, for example, scanning electron mi-

croscopes, atomic force microscopes, or X-rays)

2D-DIC and stereo-DIC

Single DIC system: One camera for 2D-DIC and two cameras

for stereo-DIC

Standard machine vision cameras and optical lenses (no images from, for example,

microscopes, stereo microscopes, or high-speed cameras)

Local, subset-based DIC algorithms (as opposed to global algorithms)

The speed of the mechanical test, i.e. quasi-static versus dynamic, is not speciﬁcally

scoped in this guide, as none of the guidelines presented here are limited to a certain

range of grip velocities or strain rates.

Volumetric DIC (DVC) and other image processing techniques such as image stitching, photogram-

metry data alignment, point tracking, object tracking, etc. are not discussed.

See footnote 1 on page 1.

The use of multiple cameras or multiple systems covering diﬀerent regions-of-interest of a test

piece, e.g. around a cylinder, is not discussed.

There are some caveats to the applicability of this guide to dynamic tests. First, this guide

is limited to standard machine vision cameras and excludes high-speed cameras, due to additional

hardware complexity. High-speed cameras are deﬁned here as cameras that record data in a burst to

a RAM buﬀer on local camera memory that must be downloaded afterward. Second, special care and

attention is often required when selecting a DIC pattern for dynamic tests, and this guide does not

provide any guidelines, tips, or recommendations regarding this topic. Third, synchronization — both

between cameras for stereo-DIC and between the camera(s) and other data of interest (e.g. force) —

is often more complex for dynamic tests, but this guide only discusses the most basic need for camera

synchronization for stereo-DIC. These and other features of dynamics tests are all beyond the scope

of the current edition of this guide.

2 — Design of DIC

Measurements

2.1 Measurement Requirements

Before conducting DIC measurements, clearly deﬁne the expectations and requirements

of the mechanical test, and the objectives of the DIC measurements. The limits chosen

here will be used later to assess if the analyzed results are within, approaching, or past

the limits for which the DIC measurements were designed.

2.1.1 Quantity-of-Interest

Select the quantity-of-interest (QOI) such as shape, displacement, velocity, accelera-

tion, strain, strain-rate, etc.

2.1.2 Region-of-Interest

Select the region-of-interest (ROI) of the test piece, and determine the expected motion

and/or deformation of this region. The ROI may be a speciﬁc portion of the entire

test piece (e.g. the gauge length and exposed end tabs of a uniaxial test piece).

2.1.3 Field-of-View

Determine the required ﬁeld-of-view (FOV) based on the ROI of the test piece and

expected motion and/or deformation of the test piece.

Recommendation 2.1

Typically, the ROI of the test piece should almost ﬁll the FOV to optimize the

spatial resolution, while still remaining in the FOV throughout the test. For

stereo-DIC, where the FOV is not the same in each camera, the eﬀective FOV is

the common FOV that is captured in both cameras, i.e. the portion of projected

images to each camera of the same region of space. See Sec. 2.2.2 for more

information about designing a camera mounting system to obtain the desired

FOV.

2.1.4 Position Envelope for Hardware

Estimate the potential position envelope for cameras, mounting hardware, and lights

to determine feasible stand-oﬀ distance (SOD) and location. Determine what size

and type of calibration target will be used (e.g. front-lit or back-lit) and how the

mechanical test setup will need to be modiﬁed in order to calibrate the optical system

(Sec. 3.2.2.2). Select (and purchase or fabricate if necessary) appropriate equipment

for camera support structure (Sec. 2.2.2).

Tip 2.1

For more information on the relationship between the FOV, lens focal length,

camera sensor size, and SOD, see Appendix B.

Recommendation 2.2

Consider adding a stationary backdrop behind the test specimen, to prevent any

people or objects moving behind the test specimen from adversely aﬀecting the

images.

2.1.5 2D-DIC vs Stereo-DIC

Determine if 2D-DIC or stereo-DIC will be used.

Caution 2.1

For 2D-DIC, the test piece is assumed to be planar, to remain planar throughout

the test, to be perpendicular to the camera optical axis,

and to maintain constant

SOD throughout the test. Any inadvertent out-of-plane motion (i.e. due to test

piece thinning or buckling, rotations or translations induced by misaligned grips,

etc.) will cause errors in 2D-DIC [13, 43, 44].

Recommendation

Stereo-DIC is strongly recommended over 2D-DIC for all tests if possible, even

tests in which a nominally planar test piece undergoes nominally planar defor-

mation. 2D-DIC is recommended only if the geometry of the test setup cannot

accommodate two cameras/lenses (i.e. when two cameras cannot physically ﬁt

into the position envelope available given the load frame and other equipment

placement).

b,c

If the intrinsic and extrinsic parameters of a 2D, single camera system are calibrated using

a calibration target as described in Sec. 3.2, then an out-of-plane tilt of the test piece can be

determined and corrected. However, this is an advanced topic that is outside the scope of the

current edition of this guide.

Lack of availability of two cameras/lenses due to cost can also prevent the use of stereo-

DIC. However, this typically only applies to high-speed or ultra-high-speed cameras, which

are outside the scope of the current edition of this guide, and not to standard machine-vision

cameras.

2D-DIC may also be required when testing highly porous foams, because the natural pattern

created by the pores can appear diﬀerently in the left and right cameras (due to diﬀerent

perspectives, lighting, and shadows of each camera looking “into” the pores), and because

applying a DIC pattern to the surface of the foam can be diﬃcult. However, speciﬁc details

about foam materials are beyond the scope of the current edition of this guide.

Recommendation 2.3

If 2D-DIC must be used, estimate the expected out-of-plane motion/deformation

of the test piece during the test (due to test piece thinning, for example) and the

corresponding error of in-plane measurements as described in Sec. 5.4.3.

See Sec. 2.2.1 for recommendations on lens selection for 2D-DIC.

2.1.6 Stereo-Angle

For stereo-DIC, select the required stereo-angle.

Tip 2.2

The stereo-angle depends on geometry of the test setup and the QOI that is most

important. Smaller stereo-angles lead to better in-plane displacement accuracy,

at the cost of increased out-of-plane uncertainty. Alternatively, larger stereo-

angles lead to better out-of-plane displacement accuracy, at the cost of increased

in-plane uncertainty.

This relationship between stereo-angle and uncertainty is also aﬀected by the

focal length of the lens. Shorter focal length lenses require a larger stereo-angle

to obtain the same out-of-plane uncertainty as longer focal length lenses.

The stereo-angle also aﬀects the useable DOF. With smaller stereo-angles, the

test piece will remain in focus in both cameras over a larger range of out-of-plane

motions. Conversely, with larger stereo-angles, the allowable out-of-plane motion

to keep the test piece in focus is reduced.

Recommendation 2.4

Typically, the stereo-angle should be between approximately 15–35 degrees [23].

To reduce out-of-plane uncertainty and maximize useable DOF, short focal length

lenses (8-12 mm) should have a minimum angle of 35 degrees and mid-range focal

length lenses (17 mm) should have a minimum angle of 25 degrees; lenses with a

focal length of 35 mm or longer can have a stereo-angle of 15 degrees [3].

Caution 2.2

Experience has shown that large stereo-angles (greater than approximately 35

degrees) may lead to diﬃculties in cross-correlation between the two cameras,

due to large perspective diﬀerences in the images from each camera, especially

with wide angle lenses.

2.1.7 Depth-of-Field

For stereo-DIC, determine the required depth-of-ﬁeld (DOF) so that the entire ROI of

the test piece remains in focus during the entire test, taking into account the expected

out-of-plane motion, and the stereo-angle of the cameras.

Tip 2.3

For 2D-DIC, the test piece is assumed to be planar and to remain planar with

constant SOD. Therefore, DOF is not a large factor in the design of a 2D-DIC

setup. However, having suﬃcient DOF helps ensure that the images will be in

focus when the test piece is inserted into the load frame, and reduces sensitivity

to alignment of the test piece and load frame with the optical axis of the imaging

system. Additionally, having suﬃcient DOF will help ensure that focus will be

maintained even during unexpected out-of-plane motion and/or deformation.

2.1.8 Spatial Gradients

Estimate the expected spatial gradients in the QOI. This will determine the required

spatial resolution of the DIC system, which is a function of measurement design pa-

rameters such as camera resolution and FOV, and DIC processing parameters such as

subset size and step size.

Tip 2.4

If the spatial gradients of the QOI are higher than the maximum gradients that

the DIC system can resolve, consider increasing the magniﬁcation of the optical

system (i.e. increasing the image scale) by (1) using a camera with higher image

resolution or by (2) reducing the ROI of the test piece to be a smaller portion of

the test piece.

These tips assume that the spatial resolution of the DIC system

is camera limited, meaning that increasing the number of pixels across the ROI

directly improves the spatial resolution of the DIC system. However, at high

magniﬁcation (small FOVs) or high image resolution, the system may be lens-

limited, meaning that further increase in magniﬁcation or image resolution will

not improve the spatial resolution.

Alternatively, two DIC systems could be set up, one at a lower magniﬁcation and with a

larger FOV and larger DIC pattern features, to capture the overall motion and deformation of

the test piece, and a second system at a higher magniﬁcation focused on a smaller ROI of the

test piece with smaller DIC pattern features, to capture a localized region of sharp gradients.

This advanced topic, however, is outside the scope of the current edition of this guide.

2.1.9 Noise-Floor

Determine the acceptable noise-ﬂoor for all QOIs. Justify and document the criteria

used to establish this acceptable noise-ﬂoor.

Tip 2.5

This threshold of an acceptable noise-ﬂoor is application-speciﬁc, and is often de-

termined by a subject matter expert. The noise-ﬂoor can be evaluated during the

design of the measurement to aid in the selection of diﬀerent DIC hardware (i.e.

camera and lens, patterning technique, lighting, etc.) and processing parameters

(i.e. subset size, step size, etc.). See Sec. 5.4 for more information.

2.1.10 Frame Rate

Determine the desired frame rate.

Tip 2.6

There are several factors to consider when determining the desired frame rate,

listed here in order of importance:

1. The most important factor in determining an appropriate measurement rate

is the desired temporal resolution of the QOIs. Therefore, the frame rate for

DIC measurements is chosen to be commensurate with the highest expected

rate of variation of any QOIs. Because temporal resolution requirements are

application-speciﬁc (see examples below), no general guidelines for frame

rate and/or number of images to acquire during the test are given here.

As an example, if the goal of the DIC measurements is to capture the yield

point of a metal in a tensile test, then the temporal resolution requirements

are a function of the number of frames required to adequately capture the

elastic-plastic transition. Alternatively, if the goal is to determine the max-

imum strain before necking of a ductile material, the frame rate could be

much slower. As a third example, if the test piece is cyclically loaded,

the minimum frame rate is determined by Shannon’s sampling theorem for

noisy signals, i.e. the imaging frame rate must be about 10–15 times the

frequency of oscillation.

2. A second, minor consideration when selecting the frame rate is the amount

of displacement between frames. If the displacement between frames is

large, DIC algorithms may fail to locate the subset position in the de-

formed images. However, most DIC software allows for an initial guess that

is chosen by the user, which usually successfully compensates for large dis-

placements between images. The maximum displacement between frames

that can be captured is software-speciﬁc; however, a reasonable value (as a

starting point) is approximately one subset size. For example, if the subset

size is 25 px and the image scale is 20 px·mm

-1

, then the maximum dis-

placement between two sequential frames should be less than approximately

1.25 mm. If the velocity of the test piece is approximately 1 mm s

-1

, then

the minimum frame rate should be approximately 0.8 Hz.

3. A third, minor consideration is the amount of data collected during the

mechanical test. Gigabytes of data can quickly be accumulated during DIC

measurements, so good data management is essential.

2.1.11 Exposure Time

Determine the maximum allowable exposure to limit motion blur.

Tip 2.7

The most conservative estimate for the maximum allowable test piece motion over

the course of the exposure time is the noise-ﬂoor (Sec. 2.1.9) of the displacement

measurements. For typical DIC setups, this threshold is around 0.01 pixels. In

some ﬁelds such as machine vision, a threshold of 0.1 – 0.3 pixels is typically used,

while dynamic modal tests often accept as much as 3 pixels of motion during the

exposure time [45].

The displacement, in pixels, per exposure is calculated as:

Displacement per Exposure [px] =



Velocity

i



Image Scale

i

· (Exposure Time [s])

(2.1)

The recommended thresholds for motion blur given here in terms of pixels assume an ideal

pattern feature size of 3–5 pixels (Sec. 2.3.2). An alternative view is that the motion blur should

be below a percentage of the mean feature size. For example, if a pattern is imaged with a

1 MP camera and the feature size is 3 pixels, then motion blur of 0.3 pixels corresponds to 10 %

of the feature size. If the same pattern is imaged with a 16 MP camera so that the feature size

is now 12 pixels, the same amount of 10 % blur corresponds to 1.2 pixels. If the subset size

is correspondingly increased in the second case, the blur should aﬀect the displacement results

similarly.

Tip 2.8

While the exposure time is determined independently from the frame rate, the

exposure time cannot be larger than the inverse of the frame rate (Sec. 2.1.10).

2.1.12 Synchronization and Triggering

Determine how the DIC images will be synchronized to other measurements of interest,

such as applied force or displacement, strain gauges, thermocouples, etc. Determine

how all data acquisitions will be triggered at the start of the mechanical test.

The method used to synchronize DIC images with other measurements of interest is dependent on

the speciﬁc hardware and software of the mechanical test; therefore, no further information is given

in this guide.

2.2 Equipment and Hardware

2.2.1 Camera and Lens Selection

Select a camera and lens

pair to obtain the desired FOV, DOF, SOD, spatial resolu-

tion, temporal resolution and noise-ﬂoor determined in Sec. 2.1.

Tip 2.9

FOV, SOD, and DOF are all intertwined and must be selected together. Cam-

eras and lenses cannot be selected independently, due to the combined detector

size and lens eﬀect on the resulting image scale. For more information on the

relationship between the FOV, lens focal length, camera sensor size, and SOD,

see Appendix B. Additionally, more information on camera and lens selection is

found in [21, 22].

Tip 2.10

In most cases, experience is necessary to determine if a camera (e.g. noise level

and dynamic range) or lens (e.g. distortions and resolution) is of suﬃcient quality

for DIC. Some qualiﬁcation or veriﬁcation of new hardware is recommended, by

characterizing the baseline noise-ﬂoor of DIC results with the new hardware.

Typically, this is done by DIC vendors for any hardware they provide, but DIC

practitioners can verify the results, or evaluate independent hardware, by using

the procedures outlined in Sec. 5.4.2.

Recommendation 2.5

Typically, machine-vision, monochromatic cameras with nearly square pixels are

used for DIC.

Also, sensors with global shutters (in which the data is read from

all pixels simultaneously) are recommended over sensors with rolling shutters (in

which the data is read from the sensor row by row).

The use of color cameras or the use of cameras that have pixels that are not square is

possible; however, the use of these types of cameras requires complex analysis and is an advanced

topic beyond the scope of the current edition of this guide.

Caution 2.3

Imaging systems that automatically adjust components of the lens and/or camera,

such as auto-focus of the lens, or apertures that open/close with each image

acquisition, are not appropriate for DIC measurements and should be avoided.

Some lenses have anti-vibration features, but the eﬀects of these features on DIC measurements

have not yet been thoroughly investigated.

Caution 2.4

Some cameras (especially older digital video) have the ability to “interlace” frames

to result in a smooth video that is more pleasing to the human eye. This feature

combines every other row of the previous frame with the current frame, and is

completely inappropriate for use in DIC.

Recommendation

When using new or unfamiliar camera hardware, it is recommended to verify that

“interlacing” is not being used.

Tip 2.11

Some cameras have a physical low-pass ﬁlter element (also called an anti-aliasing

ﬁlter) adhered in front of the detector. This is more typical of single-lens reﬂex

(SLR) or digital single-lens reﬂex (DSLR) cameras than of machine vision cam-

eras. It is important to know whether or not the camera being used for DIC has

a physical low-pass ﬁlter, in particular when deciding whether or not to pre-ﬁlter

the images, as described in Sec. 5.2.2.

Tip 2.12

There are two main types of lenses used for stereo-DIC (and occasionally used for

2D-DIC), either ﬁxed focal-length lenses or zoom lenses. With a ﬁxed focal-length

lens, the FOV or image scale is adjusted by adjusting the SOD. With a zoom

lens, the FOV or image scale can be adjusted by adjusting either the SOD or the

focal length of the lens. Thus, zoom lenses can be more ﬂexible than ﬁxed focal-

length lenses. However, because of increased complexity of the optics in a zoom

lens, lens distortions are often larger for zoom lenses. Also, many (though not

all) zoom lenses do not have a way of locking the adjustment of the focal-length

ring, making them more susceptible to inadvertent changes if the camera/lens is

moved.

Recommendation 2.6

If 2D-DIC must be used, a bilateral telecentric lens is recommended to mitigate

small errors due to out-of-plane translation; out-of-plane rotations and large out-

of-plane translations, however, will still cause errors in 2D-DIC measurements.

The magnitude of out-of-plane translations for which a bi-telecentric lens can

compensate depends on the FOV.

If a telecentric lens is not available or feasible, it is recommended to use a

long focal-length lens, to maximize the SOD, and hence minimize errors caused

by out-of-plane motion.

Caution 2.5

Although a telecentric lens is recommended for 2D-DIC, telecentric lenses may

not be used for stereo-DIC.

Recommendation 2.7

Lenses with the ability to lock moving components (e.g. focus ring, aperture

ring, zoom setting (for a zoom lens)) are preferred, to reduce the likelihood of

accidentally changing these components after they have been set to the desired

position.

2.2.2 Camera and Lens Mounting

2.2.2.1 General Characteristics of Mounting System

Construct a sturdy camera and lens mounting system with the following general char-

acteristics:

Include suﬃcient degrees of freedom to allow for precise adjustment of the loca-

tion and orientation of the camera(s)/lens(es) (i.e. translation or rotation stages,

tripod adjustments, etc.).

If the camera location and/or orientation need to be adjusted for calibration

(Sec. 3.2.2.2), include the appropriate mechanisms in the mounting system (e.g.

a bar that can rotate the camera(s) or a translation stage that can translate the

camera(s)).

Lock all moving components in the mounting system after the ﬁnal position and

orientation has been determined.

Tip 2.13

As mentioned in Recommendation 2.7, lenses with the ability to lock mov-

ing components (e.g. focus ring, aperture ring, zoom setting (for a zoom

lens)) are preferred. However, if the only lens(es) available for the DIC

measurement do not have locks for the moving components, masking tape

can be used to lock the position of these adjustment rings. The rings should

be taped after the imaging system is aligned and focused, but before the

system is calibrated. Care must be taken during taping to not change the

focus or other lens settings.

For 2D-DIC, ensure the camera and lens optical axis is perpendicular to the

surface of the test piece.

Caution 2.6

See Sec. 2.1.5 for more information about the implications of out-of-plane

motion in 2D-DIC.

For stereo-DIC, mount the cameras such that the desired FOV, image ROI, and

stereo-angle are achieved.

Recommendation 2.8

Set the FOV such that the image ROI is the same in both cameras, as close

as possible given the diﬀerent perspective each camera sees.

Set the orientation of cameras with rectangular detectors (chips) so that

the long axis of the detectors are aligned with the long axis of the ROI of

the test piece (Fig. 2.1). If the ROI is short and wide, mount the camera

in landscape orientation; if the ROI is tall and narrow, mount the camera

in portrait orientation.

Set the stereo-plane perpendicular to the plane of the test piece (Fig. 2.1).

Camera orientations with compound angles (i.e. a stereo-angle between

the cameras and a tilt of the stereo-plane with respect to the test piece)

increase perspective distortions of the images, and thus are less desirable,

and should be avoided.

Test piece ROIs with a large aspect ratio should be oriented with their

long axis (opposed to their short axis) perpendicular to the stereo-plane

(Fig. 2.1). This orientation reduces large perspective diﬀerences at the

edges of the ROI in each camera.

This guideline only applies to nominally planar test pieces imaged with a single pair

of stereo cameras. More complicated test piece geometries, DIC systems with more than

two cameras, and multi-system DIC measurements are outside the scope of the current

edition of this guide.

For stereo-DIC, mount both cameras rigidly together to avoid relative camera

motion.

See Sec. 2.2.2.2 for more information on common types of mounting

systems.

Caution 2.7

Any relative motion of one camera with respect to the second camera will

induce errors in DIC measurements.

If relative motion occurs, the camera

system should be recalibrated.

To avoid this problem, rigid mounting is

critical!

If both cameras move together rigidly with respect to the test piece, only rigid-body

DIC displacements are aﬀected. For most applications where rigid-body motion is not

important (e.g. strains are the QOI), this rigid-body displacement error is inconsequen-

tial.

Rigid-body motion of the stereo-camera pair can be corrected in post-processing if

there is a ﬁxed reference point somewhere in the FOV. However, correcting for relative

motion of one camera with respect to the second camera requires adjusting the extrinsic

parameters of the calibration (Sec. 3.2). Some DIC software packages oﬀer a “calibration

correction” that corrects the extrinsic parameters of a stereo-camera system based on

certain assumptions and minimization of the epipolar error (Sec. 3.3.2.2). However, this

type of correction is beyond the scope of the current edition of this guide.

Test setups in which the cameras cannot be practically rigidly-mounted together, e.g. large-scale

tests with large camera separation that require individual tripods for each camera, are not discussed

in this edition of the guide.

detector short axis

detector long axis

test piece

long axis

test piece

short axis

stereo-plane

detector short axis

test piece

short axis

stereo-plane

Top View

Side View

test piece

long axis

detector

long axis

stereo-plane

Isometric View

Figure 2.1: Schematics depicting

the recommended orientations of

the camera detectors in a stereo-

camera pair with respect to the

test piece. In this schematic, the

long and short axes of the test

piece ROI are assumed to align

with the long and short axes of the

test piece itself.

Mount the combined camera/lens system near its center of mass. If either the

lens or the camera is substantially more massive than the other, mount to the

more massive element. Consider mounting at two places along the optical axis

instead of just one, to minimize the lever-arm eﬀect.

Recommendation 2.9

Many commercial cameras, lenses, and mounts are designed to be used

with the optical axis nearly horizontal. If the orientation of the optical axis

is vertical, then verify the mounting, and reinforce it as needed to ensure

the mount is eﬀectively rigid in this orientation. Additionally, verify that

the lens performs properly in this orientation, and that the focus or other

settings do not drift.

Caution 2.8

Camera and lens systems that are not well balanced on their mounting are

more likely to drift or become misaligned, thereby inducing errors into the

DIC measurements.

Stabilize and strain relieve camera cables to prevent the cables from pulling on

the cameras or transferring ambient vibrations to the camera system. If the

camera(s) will be moved for calibration (Sec. 3.2.2.2), ensure there is enough

slack in the cables to accommodate the camera repositioning.

Ensure that the camera support structure is stable. If necessary, add weights

(e.g. sand bags) to tripods or other footing to prevent motion of the camera

support structure.

Minimize vibrations being transferred to the cameras.

Caution 2.9

Any vibrations that are transferred to the cameras will directly increase the

noise-ﬂoor of the DIC measurements. The amount of time and eﬀort spent

on minimizing vibrations is directly commensurate with the magnitude of

the vibrations and the desired precision of the DIC measurements.

Tip 2.14

Some vibrations — but not all — can be detected by the human eye by

watching a live image stream, especially if the image is zoomed in so that

individual pixels are visible. Camera and lens vibrations are more visible

at larger SODs, and less visible for short SODs.

Caution 2.10

A vibrating stereo-DIC system can result in temporal changes in the ori-

entation between the cameras that will invalidate the calibration, even if

vibrations are not visible to the human eye.

Recommendation 2.10

To reduce the eﬀects of vibrations, the following precautions are recom-

mended:

– Ensure the cameras and mounting system are not in direct contact

with any vibrating components (i.e. fan, compressor, hydraulics, test

machine, lights, etc.) [20]

– Verify there are no vibrations being transferred through the ﬂoor.

(Note that vibrations could come from equipment in other rooms in

the building.)

– If vibrations are being transferred to the cameras, reinforce the mount-

ing system and/or add damping.

– If the DIC measurement and test setup can accommodate diﬀerent

SODs and lenses of diﬀerent focal lengths, use a shorter SOD and

shorter focal-length lens.

Tip 2.15

In stereo-DIC, the epipolar error is a good metric for indicating possible drift,

misalignment, or vibrations in the camera/lens systems. For example, if the

epipolar error of a series of static images was low immediately after calibrating

the stereo system, but then increases over time, this could indicate drift of one or

both imaging systems. If the epipolar error is cyclic over time, this could indicate

vibrations aﬀecting the imaging systems.

2.2.2.2 Types of Mounting Systems

There are many types of mounting systems for camera(s)/lens(es) that are appropriate

for DIC, and the selection of a mounting system depends on the mechanical test setup

and components available in each laboratory. A standard mounting system, appropri-

ate for a large range of mechanical test setups, is often available with the purchase of

commercial, turn-key DIC systems. Alternatively, custom mounting systems can be

built from commercially-available products. For mechanical test setups with compli-

cated geometry, restricted access, or uncommon camera(s) and/or lens(es), mounting

system components may need to be specially designed and fabricated. Some common

types of commercially-available systems include, but are not limited to:

Standard optical hardware, such as 1-1/2 inch (38 mm) posts and associated

mounting hardware, bolted to an optical table.

Sturdy tripods. For stereo-DIC, a single bar is either mounted at each end to a

tripod or mounted to a single tripod at the center of the bar. The two cameras are

then mounted to the bar. In this way, the cameras are mounted rigidly together,

and not on independent tripods.

Refer to footnote 9 on page 14.

Studio stands. Camera mounting with studio stands is similar as mounting with

tripods, but studio stands have two advantages. First, their base is weighted,

decreasing the likelihood that sandbags or other weights will be necessary to

stabilize the system. Second, they are designed with several lockable degrees of

freedom, allowing for easy adjustment of camera location and orientation.

Systems of bars pre-fabricated to varying lengths and associated assembly and

mounting hardware.

2.2.3 Aperture

Select the aperture on the lens to obtain desired DOF (Sec. 2.1.7). For stereo-DIC,

the aperture should be the same in both cameras (as close as possible).

Tip 2.16

In addition to governing the DOF, the aperture of the lens also governs how

much light enters the optical system. However, typically the aperture is chosen

based on the desired/required DOF, and external light (Sec. 2.2.4) and exposure

(Sec. 2.1.11) are adjusted in order to limit motion blur and obtain suﬃcient

contrast.

Tip 2.17

The smaller the aperture (the larger the f-stop number), the larger the DOF.

Caution 2.11

Diﬀraction may become problematic at small apertures, while optical aberrations

are accentuated by large apertures.

Recommendation

Moderate lens apertures are recommended, to avoid accentuated lens distortions

or diﬀraction limits at extreme apertures [41]. The recommended aperture size

or f-stop number is dependent on the lens and the application, but typically a

value in the range of f/5.6–f/11 is recommended.

2.2.4 Lighting and Exposure

Given a pre-determined aperture (Sec. 2.2.3), select lighting and an exposure time (less

than to equal to the maximum allowable exposure time, see Sec. 2.1.11) to have suﬃ-

cient contrast between the lightest (white) and the darkest (black) regions of the DIC

pattern. The contrast should be uniform over the entire ROI of the image, approxi-

mately the same in both cameras (for stereo-DIC), and constant in time. For standard

stereo-DIC setups, the exposure time should also be the same for both cameras.

The exposure time is not strictly required to be the same in both cameras of a stereo-DIC setup.

The exposure time may be adjusted for each camera independently to account for diﬀerent sensitivities

2.2.4.1 Type of Lights

Tip 2.18

In some cases (e.g. slow, quasi-static tests and moderate aperture), room lighting

is suﬃcient. However, most of the time, additional lighting is required to have

good contrast for a given aperture and exposure time. In these cases, white light

or light of any wavelength or band of wavelengths will work.

(It is recommended,

though, to avoid lighting that has signiﬁcant intensity in the infrared range, as

this may increase the temperature of the test piece, and thus change the behavior

of the test piece.) The main requirement for lighting is that it is uniform and

constant, both across the FOV and in time.

Speciﬁc wavelengths of light can be advantageous over white light for some DIC measure-

ments, but this is an advanced topic outside the scope of the current edition of this guide.

Recommendation 2.11

Cross-polarized light [12] or diﬀuse light (instead of focused or spot light) are

recommended to reduce glare caused by specular reﬂections.

See Sec. 2.3.2.5 for

more information about specular reﬂections.

Cross-polarized light is especially useful for curved surfaces or test pieces that are antici-

pated to undergo large rotations, but these topics are outside the scope of the current edition

of this guide.

Caution 2.12

Some lights may ﬂicker (i.e. change intensity) at the same frequency as the

alternating-current (AC) electrical supply (typically 50–60 Hz). Similarly, for

some LED lights, the intensity of the light is controlled by varying the duty cycle

of the light, again typically at 50 Hz. In either case, if the imaging frequency

(i.e. frame rate) is close to or faster than the AC electrical supply frequency or

the duty cycle frequency, then the intensity of the light (and thus the contrast of

the images) can vary between images.

2.2.4.2 Light Mounting

Tip 2.19

Often, diﬀerent lighting and/or exposure is required to have good contrast for

the test piece versus a calibration target (see Sec. 3.2 for information on DIC

system calibration). Adjusting the light intensity, position, and/or the exposure

time for calibration purposes is a common procedure and is acceptable as long

as the camera, lens, and mounting are not disturbed (see Sec. 3.2.2.3), and the

adjustments are reversed before the DIC measurements are made.

of the detectors, for instance, as long as each exposure time is short enough to limit motion blur as

described in Sec. 2.1.11. However, this is an advanced topic outside the scope of the current edition

of this guide.

Recommendation 2.12

Numerous vendors supply lights that are integrated robustly onto the camera

mounting system; these systems are designed to allow adjustment of the lights

without disturbing the cameras. Alternatively, lights can be mounted on a sep-

arate frame than the cameras, or remote lighting control can be used, to reduce

the possibility of unintentional camera motion when adjusting lights.

2.2.4.3 Contrast, Intensity, and Gain

Recommendation 2.13

Because DIC metrological properties rely highly on image gradients (and thus

contrast), the better the contrast is (without the image being overexposed or

underexposed), the less noisy the DIC results are. For an 8-bit camera, the

minimum contrast to have a displacement noise-ﬂoor of around 0.005 pixels is

approximately 20% (50 grey-level counts between the light and dark features)

[34], though contrast of at least 50% (130 counts) is typically preferred.

More advanced users may optimize the contrast beyond simply the diﬀerence between the

light (white) and dark (black) intensities. A well-spread distribution of intensities between

these limiting white and black values behaves better than a bi-modal distribution. It can be

advantageous to use only the lower portion of the dynamic range of the camera detector, as long

as there is still suﬃcient contrast, because camera noise typically scales with intensity (though

this can be camera-speciﬁc) [1].

Tip 2.20

There can be suﬃcient contrast in the image for DIC, even if the images “look”

dark to the eye. This is especially true for camera detectors having larger dynamic

ranges.

Caution 2.13

Contrast may change during the test, as the test piece is moved and/or deformed.

Therefore, ensure there is suﬃcient contrast both at the beginning and throughout

the duration of the test. Pre-testing extra test pieces may be required to conﬁrm

lighting and contrast throughout the test.

Recommendation

If the mean contrast in the image changes over time during the test, the zero-mean

normalized sum of square diﬀerence (ZNSSD) matching criterion is recommended

to compensate for contrast changes.

Caution 2.14

Ensure no regions-of-interest of the image are overexposed (i.e. the intensity at

every pixel should be less than the maximum of the camera) or underexposed (i.e.

the intensity at every pixel should be greater than the minimum of the camera),

and there is no glare in the ROI. These conditions should be true both initially

and as the test piece is translated and rotated within the expected 3D volume of

motion and/or deformation [24]. In stereo-DIC, check both images for glare, as

glare could appear in only one of the two cameras. Keep in mind that glare can

manifest as either points or as lines.

Caution 2.15

Do not increase the gain (sometimes referred to as the exposure index or the “ISO

setting”) of the camera(s) — the conversion between the number of electrons

recorded by the detector and the number of counts in the gray-level intensity —

in an attempt to increase the contrast or intensity. Increasing the gain increases

camera noise, with no beneﬁt for DIC.

2.2.5 Hardware Heating

Caution 2.16

Almost all cameras and lights become hotter than room temperature when run

continuously, even “cool” LED lights. Hardware heating can negatively impact

DIC measurements in several ways, including but not limited to:

Changing the sizes and positions of the camera detector(s) and lens(es) due

to thermal expansion of the components of the camera(s) and lens(es).

Heating of the mounting structure, which, in stereo-DIC, can result in a

change in the relative positions of the cameras during testing that negates

the calibration.

Inducing convective air currents (known colloquially as “heat waves”) that

refract light between the test piece and the imaging system.

Heated test pieces may also induce heat waves, but heated test pieces are excluded from

the scope of the current edition of this guide, and are not discussed further.

Recommendation 2.14

To mitigate eﬀects of thermal expansion of the camera(s), lens(es) and mounting

structure, cameras should be turned on and operated at the target frame rate

(Sec. 2.1.10) until they have reached a stable operating temperature. Calibration

images and DIC measurement images should only be acquired after the cameras

have reached thermal equilibrium. See Sec. 3.1.3 for more information on warming

up the cameras.

Tip 2.21

Even small temperature changes can cause heat waves between the test piece and

the imaging system, which refract light and cause errors in DIC measurements.

These errors manifest as spatially- and temporally-varying “ﬁngers” in the dis-

placement and strain ﬁelds that, when animated in a movie, can look similar to

ﬂames in a ﬁre [9].

Recommendation 2.15

Because errors caused by heat waves are not easily ﬁltered in post-processing [9],

it is strongly recommended to minimize heat waves in the test setup before images

are acquired, using one or more of the following preventative steps. Similar to

camera vibrations (Sec. 2.2.2), the amount of time and eﬀort spent on minimizing

heat waves is directly commensurate with the magnitude of the errors caused by

heat waves, and the desired precision of the DIC measurements.

Mount lights above and behind the camera(s) if possible. Avoid mount-

ing lights between the camera(s) and the test piece. In particular, avoid

mounting lights below the camera/test piece plane.

If lights are the source of heat waves and the test duration is short, keep

lights oﬀ as much as possible to prevent them from heating up, and turn

them on only for the test duration. If the test duration is long, strobe the

lights on only when images are being acquired, or add a fan onto the lights

to cool them and homogenize the air temperature.

If heat waves are caused by the camera(s), cool the camera(s) with a heat

sink or a fan. Alternatively, place an air knife in front of the cameras, to

homogenize the air between the cameras and the test piece without blowing

air directly onto the camera(s).

Caution 2.17

Beware of inducing camera motion by blowing air onto the camera(s) and/or

transferring vibrations from a fan to the camera mounting structure. If using a

fan to cool lights or camera(s), ensure that the reduction in errors due to reducing

heat waves is more impactful than any increase in errors due to camera motion.

2.3 DIC Pattern

2.3.1 Type of DIC Patterns

One fundamental assumption of DIC is that the motion and deformation of the pattern

that is imaged exactly replicates the underlying test piece motion and deformation.

Sometimes, images of the surface of the test piece itself have a suﬃcient natural pattern

that is adequate for DIC, and no artiﬁcial pattern needs to be applied.

Tip 2.22

As a ﬁrst step, the test piece surface can be imaged, and the natural pattern can

be evaluated to ascertain if it has the characteristics described in Sec. 2.3.2. If

so, no applied DIC pattern is necessary.

Caution 2.18

Be cognizant of any surface coating that may be on the surface of the test piece,

i.e. brittle mill scale on steel, brittle oxides, added coatings such as zinc coating

on steel, etc. Such surface coatings may or may not move with or behave like

the underlying material. Ensure that the pattern being imaged on the surface

reﬂects the deformation of interest of the bulk material underneath.

Most of the time, a pattern must be applied to the test piece surface. Typically,

though not exclusively, applied patterns consist of roughly circular “speckles” of a

(preferably) uniform size but random locations.

Recommendation 2.16

The quality of a DIC pattern is often evaluated by manual, visual inspection of

the images, where the DIC practitioner looks for the characteristics described

below. More quantitative evaluation of a DIC pattern is typically not necessary

outside of research activities, but can consist of metrics such as image gradients

to evaluate contrast, and other image morphology methods to evaluate feature

edges, shapes, sizes, and distribution [4, 7].

2.3.2 General Characteristics of DIC Patterns

Both natural and applied patterns should have the following general characteristics:

2.3.2.1 Size

The optimum pattern feature size is 3–5 pixels [33]. This guideline applies to both the

light (white) and the dark (black) features.

Tip 2.23

There are many deﬁnitions of feature size and many methods of determining

feature size. However, a rough, manual estimate, in which the DIC practitioner

zooms in on an image and approximates the feature size by eye, is typically

suﬃcient.

More advanced “optimized” pattern designs are outside the scope of the current edition of this

guide, but information can be found in [2, 5, 11]. Additionally, some DIC manufacturers prefer a

regular (not random) pattern, but these are also not covered here.

Caution 2.19

Pattern features that are smaller than 3 pixels risk being aliased and adding error

to DIC results [32, 40]. This error is more pronounced when displacements are

small, due to a lower signal-to-noise ratio. In the case of a compression test,

features that started at the lower end of the recommended range (e.g. 3 pixels)

may become aliased as the pattern is compressed and the features and spacing

are reduced. Features that are larger than necessary (i.e. larger than 5 pixels)

will require larger subsets (see Sec. 5.2.5) and thus will degrade spatial resolution

of displacements and strains, but will otherwise not negatively eﬀect the results

(i.e. will not add noise).

Recommendation

For many applications, if the selected patterning method results in a wide spread

of feature sizes, then it is generally preferred to use features that are larger than

optimal, to limit the number of aliased features. That is, added noise due to

aliased features is typically worse than degraded spatial resolution due to large

features.

Tip 2.24

The physical pattern feature size is determined based on the image scale. For

example, given an image scale of 20 pixel mm

-1

, a target size of 5 pixel features

translates to a physical size of (5 pixel)/(20 pixel mm

-1

) = 0.25 mm.

Note that the physical feature size required for a given DIC measurement

depends on both the FOV and the image resolution. For a given FOV and lens,

a camera with a lower image resolution (e.g. 1 MP camera) will require larger

features and spacing than a camera with a higher resolution (e.g. 12 MP camera).

Recommendation 2.17

In stereo-DIC, where the cameras are at an angle to the test piece, the image scale

(on the surface of the test piece) is not constant over the FOV of each camera. To

ensure that the smallest DIC pattern feature is not aliased at any location in the

ROI of either camera, consideration must be given to the changing image scale

across the ROI. Therefore, the location in the ROI, of either camera, where the

image scale is smallest should be found and used to deﬁne the smallest allowable

DIC pattern feature.

More advanced users may take the varying image scale into account when designing an

optimized DIC pattern. However, this advanced topic is outside the scope of the current edition

of this guide.

Tip 2.25

After the desired physical size of the DIC pattern features is calculated based

on the image scale, the desired physical size can be conﬁrmed by printing a

synthetic pattern on paper using a standard oﬃce printer (if the printer has

suﬃcient resolution for the desired pattern size) and imaging the pattern using

the same optical system as the actual test, to ensure all features are 3–5 pixels

in size.

2.3.2.2 Variation

The pattern should have suﬃcient random variation such that subsets in diﬀerent

regions of the image can be uniquely identiﬁed.

Caution 2.20

Oriented regular (anisotropic) patterns can be problematic for many DIC sys-

tems, and should be avoided unless motion in only one direction is required. For

example, a pattern based on a series of periodic lines of diﬀerent widths can result

in correlation of motion normal to the lines, but with no correlation of motion

parallel to the lines.

Recommendation

If a regular pattern (e.g. repeated printed pattern) is used, then some randomiza-

tion should be added in addition to the pattern (e.g. adding some random marker

dots to the printed pattern, or varying the width of regular periodic lines).

2.3.2.3 Density

Pattern density should be approximately 50 % (i.e. there should be approximately the

same area of light (white) and dark (black) pixels in any intended subset of the ROI

of the image) [36]. If round speckles are used, then a density closer to 25–40 % can be

expected due to the required minimum spacing between the round speckles.

2.3.2.4 Quality

Pattern quality degradation should be minimized and not permitted to result in decor-

relation during the analysis.

Tip 2.26

For natural DIC patterns, sources of pattern degradation include, but are not

limited to, signiﬁcant morphological changes and development of slip bands on

the surface of the test piece during plastic deformation.

For applied DIC patterns, sources of pattern degradation include, but are not

limited to, fading, cracking and debonding.

Note that in tests involving large deformation (e.g. several hundred percent deformation in elas-

tomers), a well-bonded applied pattern may deform so much that correlation is lost between the ﬁrst

image and an image later in the test, even if it does not debond or crack. In this case, incremental

correlation may be used. This situation of large deformation, however, is outside the scope of the

current edition of this guide and will not be discussed further.

Tip 2.27

Pretesting of extra test pieces may be required to verify the suitability of a pattern

throughout the duration of the test.

Tip 2.28

Even at strains where decorrelation does not occur, pattern degradation can result

in reduced correlation quality and increased uncertainty in the measurement [7].

2.3.2.5 Reﬂections

The pattern sheen should be matte and not glossy, to avoid glare and specular reﬂec-

tions.

Caution 2.21

Specular reﬂections can often be hidden in an otherwise good DIC pattern (i.e.

appearing as artiﬁcial bright spots in one or both of the camera images). Spec-

ular reﬂections are dependent on the orientation and position of the test piece

with respect to the light source and camera, and can change if the test piece is

rotated or translated. Additionally, in stereo-DIC, specular reﬂections often look

diﬀerent in each camera, which eﬀectively makes the DIC pattern diﬀerent and

uncorrelated in each FOV. Therefore, specular reﬂections should be avoided.

Recommendation 2.18

To reduce specular reﬂections, use cross-polarized light, or diﬀuse light, as de-

scribed in Recommendation 2.11 in Sec. 2.2.4. If specular reﬂections cannot be

suﬃciently minimized through the lighting, a photographic dulling spray can be

applied to the DIC pattern. However, if a dulling spray is used, the DIC pat-

tern should be carefully evaluated, to ensure that the spray does not degrade the

pattern.

2.3.3 Characteristics of Applied Patterns

Applied patterns, regardless of the method used to create them (i.e. painting, applying

an adhesive-backed foil or sticker, stamping or drawing with ink, applying a powder,

transfer printing, etc.), should have the following additional characteristics, which do

not necessarily apply to natural patterns.

Recommendation 2.19

For tensile tests, before applying a pattern to the test piece, mask the grip sections

so that the pattern is not applied to the areas of the test piece that will be gripped

in the load frame. This will help increase grip force, reduce likelihood of the test

piece slipping within the grips, and prevent clogging of the grips with the pattern

material (e.g. paint).

2.3.3.1 Compliance

The applied pattern should be thin and compliant relative to the test piece, such that

it does not change the test piece behavior being measured during the test.

Caution 2.22

If the applied pattern is thick and/or stiﬀ compared to the test piece, the DIC

measurements based on images of the pattern may reﬂect the deformation of

the applied pattern, rather than the deformation of the underlying material of

interest.

2.3.3.2 Bonding

There should be good bonding between the test piece and the applied pattern.

Recommendation 2.20

Before applying a DIC pattern, clean the test piece to promote good bonding

between test piece and pattern. For example, for common metals (e.g. steel and

aluminum), acetone can be used ﬁrst to remove grease, cutting ﬂuid, ink, etc.

However, acetone leaves a residue after evaporating, so test pieces cleaned with

acetone should always be subsequently cleaned with a solvent that does not leave

a residue, such as isopropanol.

Additionally, if the surface is very smooth, consider roughing it with sand

paper to promote adhesion of the applied pattern to the test piece surface, if

such treatment will not alter the test piece properties.

Caution 2.23

Debonding of an applied pattern can be an insidious problem. In some cases, an

applied pattern may locally debond from the test piece, yet remain intact and

continue to deform independent of the test piece. In these cases, the fact that

the pattern debonded may not be obvious.

Recommendation

Inspect the test piece and applied pattern closely after the test, to look for any

indications or evidence of pattern debonding.

2.3.3.3 Fidelity

The applied pattern should move and deform conformally with the test piece surface.

Tip 2.29

For paint-based patterns, the ductility of the paint should be aligned with the

expected deformation. That is, for test pieces that are expected to undergo

large deformation, the paint should be as ductile as possible, so that it stretches

with high ﬁdelity with the underlying test piece without cracking or debonding.

To accomplish this, the test should be carried out immediately after painting.

Alternatively, individual features (e.g. speckles) can be placed directly on the

test piece, without a base coat of paint.

On the other hand, if the test piece is brittle and observation of crack prop-

agation is important, the paint should be as brittle as possible, while still not

debonding or cracking independent of the test piece, so that the paint cracks at

the same time as the test piece. In this case, the paint should be allowed to fully

cure, and can even be baked (if baking does not alter the test piece properties)

to make it more brittle. Alternatively, individual features (e.g. speckles) can be

placed directly on the test piece, without a base coat of paint, so that cracking

of the test piece can be observed directly.

If no base coat is used, and individual features are placed directly on the

test piece, ensure that there is suﬃcient contrast and that there are no specular

reﬂections (see Sec. 2.3.2.5).

Caution 2.24

Sometimes, inexperienced DIC users think that an interference-based laser speckle

pattern would produce a good pattern for DIC. However, this type of pattern will

decorrelate from the motion of the test specimen for large displacements, and is

not recommended for DIC.

There are limited conditions in which a laser speckle pattern can be used for DIC, but this

is an advanced topic that is outside the scope of the current edition of this Guide.

2.3.3.4 Thickness

The pattern should be of uniform thickness.

Caution 2.25

In stereo-DIC, a pattern with a rough surface can result in the same portion of

the pattern appearing substantially diﬀerent between the left and right camera

images, which can hinder cross-correlation between the two images. In 2D-DIC, a

pattern with areas of diﬀerent thicknesses can result in artiﬁcial strain gradients

across the transitions between the areas of diﬀerent thicknesses.

2.3.4 Patterning Techniques

There are many diﬀerent techniques available to create appropriate patterns for DIC,

such as stencils, stamps, incomplete layers of paint (either using a commercial spray

paint can, or using an air brush), and printer toner or other ﬁne powders, to name

a few. Patterning techniques are limited only by the imagination. Often, patterning

techniques are learned through on-the-job training by more experienced DIC practi-

tioners, through training by a vendor when a new DIC system is purchased, or through

classes taught by subject-matter experts.

Due to the immense variety and nuances of

patterning techniques, no guidelines are given here concerning the execution of speciﬁc

patterning techniques.

Tip 2.30

In order to facilitate the design of DIC measurements, creation of a table of

patterning techniques and the resulting pattern size is recommended for each

DIC practitioner or laboratory. Additionally, a library of physical chips with

patterns created with diﬀerent techniques is helpful when designing a new DIC

measurement, because the size and contrast of diﬀerent patterns can be quickly

evaluated with the preliminary camera, lens, and lighting setup.

Tip 2.31

Once a patterning technique has been selected, pattern a scrap test piece and

image the pattern to verify that the pattern has the right size, shape, distribution

and density. Images should be taken in the test setup or in a mock setup that

uses the same cameras, lenses, SOD, stereo-angle, etc. as the actual setup. The

scrap test piece should be the same material as the actual test piece of interest,

as patterning techniques can produce diﬀerent results on diﬀerent materials. (For

example, spray-paint speckles may be larger on metal, where the paint droplets

spread out, than on cardboard or paper, where the paint droplets soak into the

material.)

iDICs oﬀers DIC classes at their annual conference. For more information, see www.idics.org.

3 — Preparation for the

Measurements

3.1 Pre-Calibration Routine

3.1.1 Review of Test Procedure

Before preparing for and executing the mechanical test with concurrent DIC measure-

ments, review the overall test procedure:

Evaluate the tentative testing procedure to be used, and ensure that no steps in

the procedure that occur after a test piece has been patterned, but before it will

be tested (e.g. gripping or assembly), will damage the pattern. Modify the testing

procedure if necessary to reduce the likelihood of scratching or contaminating the

pattern (e.g. oil on the surface of the test piece).

Ensure the mechanical load frame is properly adjusted and calibrated.

Review the time line of the test process and ensure there is adequate time for all

steps, such as warming up the cameras, warming up the load frame (if necessary),

calibrating the DIC system, reviewing the calibration, preparing and patterning

the test piece, testing the test piece, etc. Determine at what point in the test

process the test piece should be patterned (if using an applied pattern).

Ensure environmental conditions (e.g. temperature) will be stable during the

course of the DIC calibration and mechanical test.

Consider adding a stationary backdrop behind the test specimen, to prevent any

people or objects moving behind the test specimen from adversely aﬀecting the

images.

3.1.2 Cleanliness of Equipment

Ensure there is no dirt, dust, or other foreign particles (e.g. water, marks, oil, smears,

ﬁngerprints) on lens, camera detector, or calibration target.

This consideration is more important for outdoor testing, though outdoor testing is outside the

scope of this edition of the guide.

Recommendation 3.1

Keeping a clear lens ﬁlter (or linear polarizer if using cross-polarized light) on a

lens as a semi-permanent addition to the lens protects the lens and makes cleaning

easier. When lenses and cameras are not in use, lens caps and body caps should

be used to protect the equipment. Store calibration targets in protective cases to

keep them from getting dirty or damaged.

Recommendation 3.2

Image a white sheet of paper or other bright, solid background and look for

any blurred spots or smears that could indicate dirt in the optical system. The

image may need to be digitally magniﬁed until individual pixels are visible. Be-

cause the sheet of paper may not be perfectly clean itself, translate the sheet.

If the spots/smears move with the paper, then the dirt is on the paper; if the

spots/smears remain stationary, then the dirt is somewhere in the optical system.

To determine if dirt is on the lens or detector, rotate the lens. If the dirt rotates

with the lens, the dirt is on the lens; otherwise, if the dirt remains stationary

while the lens is rotated, the dirt is on the camera detector. Additionally, the

external surface of the lens and the camera detector (with the lens removed) may

be visually examined for the presence of dirt.

Tip 3.1

Often, pressurized air, such as canned air or a bellows-type or bulb-type blower,

is suﬃcient to remove dust or particles from the lens or detector. If canned air

is used, the bottle should be kept upright and not shaken, to prevent propellent

or condensate from being expelled from the can. Also, the canned air should

ﬁrst be sprayed away from the lens or detector, to ensure that no propellant or

condensate is being emitted. Another tool that can be used to remove dust or

particles is an optics cleaning brush.

To remove other contaminants such as oil, ﬁngerprints, etc., lens paper and

alcohol-based lens cleaning solution can be used to clean lenses. There are also

speciﬁc products made to clean camera detectors, which consist of a swab that

matches the camera detector size, and an appropriate cleaning ﬂuid. Do not

use denatured alcohol (also called methylated spirits) to clean a lens or detector;

denatured alcohol can contain up to 4 % water, which, while evaporating, can

bind dirt onto the lens or detector. Always follow manufacturer’s instructions for

cleaning lenses or camera detectors.

Caution 3.1

Anytime the optical system is exposed (e.g. lens caps and/or body caps are

removed and/or the lens is removed from the camera), be careful not to introduce

dirt into the optical system. Be very careful when cleaning a lens or camera

detector, as they can be easily, and irrevocably, damaged!

3.1.3 Camera Warm-Up

Turn on the camera(s) and operate them at the target frame rate (Sec. 2.1.10) to allow

them to warm up to a stable operating temperature. The camera(s) should be at a

stable operating temperature before any calibration or DIC measurement images are

acquired. Refer to the DIC vendor manual for more information for vendor-supplied

cameras.

Tip 3.2

The time required for a camera to warm up to a steady temperature depends

on the camera and laboratory environment, as well as the acquisition rate of the

images. Typically, warm-up times range from several minutes to several hours.

Before using a new camera for DIC, monitor camera temperature during the

warm-up period in the expected (or similar) laboratory environment, at the de-

sired acquisition rate, and note the time required for the temperature to plateau.

Use this warm-up time for all future DIC measurements that utilize that camera

and that acquisition rate. If the acquisition rate is changed, the warm-up time

will need to be re-computed.

Caution 3.2

If the cameras are not warmed up, errors can be introduced into DIC results

due to thermal expansion of the cameras and lenses, and due to drift induced by

thermal expansion of the camera mounts. See Sec. 2.2.5 for more information.

3.1.4 Synchronization

For stereo-DIC measurements, ensure the two cameras are synchronized to each other.

For either 2D-DIC or stereo-DIC, review the data acquisition plan, and ensure any

external signals (i.e. force, extensometers, strain gauges, etc.) are synchronized with

the DIC camera(s).

Caution 3.3

Synchronization of the cameras in stereo-DIC is critical! Delay between the two

cameras will result in errors in the DIC measurements.

Tip 3.3

Synchronization of the two cameras can be veriﬁed in many diﬀerent ways, in-

cluding:

Image a moving test piece that has a DIC pattern, correlate the images in

the DIC software, and verify that the epipolar error is acceptable based on

the DIC software documentation.

Image a strobe light set to the same frequency as the image acquisition

frequency.

Image a dynamic event and ensure the event occurs in the same frame

number in both cameras. (The speed of the dynamic event must be scaled

appropriately with the image acquisition rate.)

Measure the strobe or exposure signal from the cameras on an oscilloscope,

if a strobe or exposure signal is output by the cameras.

These signals are typically only available on high-speed cameras, which are outside the

scope of the current edition of this guide, and are not discussed further.

3.1.5 Application of the DIC Pattern

If using an applied DIC pattern (as opposed to the natural surface of the test piece),

apply the selected DIC pattern to the test piece.

Recommendation 3.3

Apply two ﬁducial marks a known distance apart on a portion of the test piece

that is within the FOV, but outside the critical ROI. Assess the uncertainty in

distance between the ﬁducials. These ﬁducials can be used to approximately

verify the camera calibration, as described Sec. 3.3.2. Other ﬁducial marks can

also be useful. For example, ﬁducial marks for the center line axis of a test piece

or center of gauge section can be used to rotate the DIC measurement results to

the test piece coordinate system.

Experience has shown that the use of certain inks, such as red ultra-ﬁne-point

permanent markers, to draw ﬁducial marks or lines before painting, can result in

bleeding of the lines through the paint in such a way that they are visible, but do

not overly degrade the pattern with respect to image correlation. Alternatively,

dotted or dashed ﬁducial marks made on top of the pattern can still be readily

detectable by manual inspection of the images, but will not degrade the pattern.

3.1.6 Pre-Calibration Review of System

Caution 3.4

This is the time to make adjustments and ﬁx any issues with the DIC mea-

surement setup, so that the best possible images are obtained. Once calibration

images are taken, very few aspects of the DIC system can be changed without

re-taking calibration images. If any adjustments are made to the optical system

hardware (cameras or lenses), then the previously acquired images must be dis-

carded and an entirely new set of calibration images must be acquired. Care and

time at this point in the test procedure can save tremendous time later on.

3.1.6.1 Position Test Piece and Cameras

Place the test piece in test frame. Position the camera(s) to obtain the desired FOV,

image ROI, and stereo-angle (for stereo-DIC). Set focus and aperture on the lens(es).

Tip 3.4

To make sure that the test piece is in the middle of the DOF, and that the focus is

constant across the ROI, start with the aperture wide open and adjust the focus

to ﬁnd the bounds of the DOF. With the aperture open, the DOF is limited; thus

it is easier to see when the test piece goes out of focus. Once the focus is set,

decrease the aperture to have the desired DOF.

3.1.6.2 Verify Optical System

Verify FOV, focus, and DOF by translating the test piece within the region of the FOV

in which it is expected to move and deform during the test.

3.1.6.3 Lock Adjustable Components

Adjust orientation of polarization ﬁlters if using cross-polarized light. Lock focal length

(for a zoom lens), focus, and aperture rings if locks are included on the lenses. Strain

relieve any loose or hanging cables.

3.1.6.4 Review Images

Review the image of the pattern using either a live image or an acquired static image.

Look carefully for:

Glare

DIC pattern that is too coarse (i.e. fewer than 3 features per intended subset

size) or too ﬁne (i.e. features that are smaller than 3 pixels)

Defects in applied pattern (e.g. scratches, smudges, foreign objects)

Out-of-focus regions of the image

Poor contrast

Non-uniform lighting (either across the FOV, in time, or between two cameras

in stereo-DIC)

Overexposed or underexposed regions

Dirt or foreign object on lens or camera detector

Vibrations or other camera motion (some of which can be detected by zooming

in on a live image and looking for non-random motion)

Recommendation 3.4

For a more thorough check of the DIC system, correlate static images of the test

piece, as correlation results can often elucidate issues that are not obvious from

visual inspection of the images. For 2D-DIC measurements, check that sequential

static images correlate. For stereo-DIC measurements, since at this point in time

the stereo system has not yet been calibrated, use a 2D-DIC software to check

that sequential images from each camera correlate.

If a certain region of the ROI in an image shows localized values of high

correlation residual, look for the cause and remedy it before moving on. Some

common sources of poor correlation results include (but are not limited to) the

bulleted list above. As a diagnostic tool, the pattern may be translated and a

couple more images may be acquired. If the region of poor correlation moves with

the test piece, then the cause is likely something on the test piece (e.g. poor DIC

pattern). If the region is at a ﬁxed pixel location, then the cause is likely dirt

or foreign object on the lens or camera detector (Sec. 3.1.2) or ﬁxed light scatter

reﬂections into a lens or camera (Sec. 2.2.4, Sec. 2.3.2.5).

Look for the presence of heat waves in the displacement ﬁelds. If signiﬁcant

heat waves are present, modify the test setup to minimize them. See Sec. 2.2.5

for more information.

3.1.6.5 Accept the DIC System

If the system is found to be acceptable, then proceed to calibration. If there are any

unsatisfactory features in the images, including but not limited to the bulleted list in

Sec. 3.1.6.4, adjust the DIC system to eliminate it. Then repeat this process iteratively

as the system is modiﬁed, until satisfactory images are obtained.

Caution 3.5

Once satisfactory images are obtained, do not modify the system, and take care

to not accidentally bump the camera(s), lens(es), or mounting system. Even the

addition or removal of lens caps can subtly change the camera position or lens

focus.

3.2 Calibration

3.2.1 Purpose of Calibration

The goal of calibration of a 2D-DIC system is to establish the image scale, i.e. the

number of pixels in the image that corresponds to a certain physical distance on the

test piece, and to correct for lens distortions.

The goal of stereo-DIC calibration is to

determine both the intrinsic camera parameters (i.e. image scale, focal length, image

center, lens distortions, etc.) as well as the extrinsic parameters of the stereo-DIC

system (i.e. stereo-angle, distance between cameras, distance from cameras to object,

etc.).

If the intrinsic and extrinsic parameters of a 2D, single camera system are calibrated, then an

out-of-plane tilt of the test piece can be determined and corrected. However, this is an advanced topic

that is outside the scope of the current edition of this guide.

Typically, both intrinsic and extrinsic parameters are calibrated simultaneously in a stereo-DIC

system. However, some software allows for calibration of the intrinsic parameters of each camera-lens

pair, and calibration of the extrinsic parameters of the stereo-system separately. However, this process

is outside the scope of the current edition of this guide.

Caution 3.6

If strains are the primary QOI, then calibration of a 2D system is often overlooked

and considered unnecessary, since strain is a unitless quantity, and an image scale

is not required to calculate strain. However, neglecting to correct lens distortions

may add error to the measured displacements and result in additional error in

the calculated strains.

Recommendation 3.5

When using 2D-DIC, assess the magnitude of lens distortions for a given optical

system, by acquiring and correlating images of a DIC pattern as it is translated

in-plane across the FOV. It is important that the translation remain strictly

perpendicular to the optical axis; otherwise, false strains due to out-of-plane

motion will be convolved with lens distortions. If errors from lens distortions

are negligible (i.e. insigniﬁcant compared to the overall noise-ﬂoor (Sec. 5.4.2)),

then 2D calibration can be omitted. If errors from lens distortions are signiﬁcant,

however, calibration is strongly recommended, to determine the intrinsic camera

parameters and correct lens distortions.

If full calibration of a 2D-DIC system to correct lens distortions is omitted,

a simpliﬁed calibration to establish the image scale is still recommended. An

approximate image scale can be computed by dividing the FOV by the camera

resolution. Alternatively, the image scale can be calculated from images of a

resolution target. It is recommended to verify the image scale in both the vertical

and horizontal directions.

3.2.2 General Calibration Steps

Caution 3.7

Before beginning the calibration process, be sure that all steps in the pre-calibration

review of the DIC system (Sec. 3.1.6) have been completed.

3.2.2.1 Select Calibration Target

Select a calibration target of an appropriate size. Consult the manual of the DIC

software for recommendations regarding the selection of an appropriate calibration

target.

Recommendation 3.6

Ideally, the calibration target should be approximately the same size as the FOV,

or slightly smaller. If no calibration target is available that is approximately

the same size as the FOV, then two other options are possible. A ﬁrst option

is to print the correctly sized target on computer paper using a standard oﬃce

printer, and glue or tape the paper calibration target to a rigid plate. In this

case, the feature spacing should be accurate to within 0.1 pixels [15]. A second

option is to use a smaller target; however, the target should not be smaller than

approximately one half of the FOV.

In this case, ensure that the features on

the calibration target are still large enough to be resolved by the imaging system

and extracted by the DIC software. Also, additional calibration images will be

required in order to have a suﬃcient number of well-extracted features in the

entire working volume of the optical system, which is discussed in Sec. 3.2.2.4.

Targets smaller than one half of the FOV may produce acceptable calibrations, but extra

precautions are required, which are beyond the scope of the current edition of this guide.

Recommendation 3.7

Scaling of DIC coordinates and displacements from pixels to physical units (e.g.

millimeters) is completely dependent on accurate and precise measurements of

the feature spacing on the calibration target. If the physical units are a critical

aspect of the measurements to be made, it is recommended to use calibration

targets that have been independently measured and are metrologically traceable

to the International System of Units (SI).

3.2.2.2 Clear Working Space

Create a clear working space in which the calibration will be performed, so that the

selected calibration target can be held, rotated, tilted, and translated as needed (re-

quirements will be diﬀerent for 2D-DIC and stereo-DIC calibration) at approximately

the same SOD as the test piece.

Recommendation 3.8

There are two strategies for creating a clear working space:

1. Remove the test piece from the test frame and move the grips back (if

necessary).

2. Move the DIC system. If moving the DIC system is necessary, then the

following are recommended:

Move the system along only a single degree of freedom, such as trans-

lating the rig backwards away from the test frame, or rotating the bar

on which the two cameras are mounted. Moving the system along two

or more degrees of freedom (e.g. rotating and translating, or translat-

ing along two directions) is also permissible, but not preferred.

If possible, calibrate the cameras in the same orientation (i.e. hor-

izontal or vertical) as they will be mounted during the test. If the

orientation is changed between the calibration and the test, the optics

inside the lens may shift slightly, changing the focus, aperture, zoom

etc.

Caution 3.8

If the DIC system is moved, it is imperative that the two stereo cameras are

moved only as a rigid pair, and that there is no relative motion between the two

cameras. Ensure that the two cameras are locked rigidly together during any

adjustment of the position of the stereo rig. Any relative motion between the two

cameras — even small changes on the micrometer scale — during calibration, or

when repositioning the cameras after calibration, will result in errors in the DIC

measurements.

Recommendation 3.9

Veriﬁcation of the calibration (see Sec. 3.3.2) is strongly recommended after re-

turning the test specimen and/or stereo system to the position to be used for

measurements, to ensure that no relative motion occurred between the two cam-

eras during the repositioning of the test specimen and/or stereo system.

3.2.2.3 Adjust Lighting

Ensure the contrast is suﬃciently large and uniform across the entire calibration target,

and that there is no glare for all desired positions and orientations of the target. These

conditions should be true for both cameras for stereo-DIC. Adjust lighting and/or

exposure time if necessary.

Tip 3.5

The lighting and exposure for the actual DIC measurement images of the DIC

pattern and for the calibration target are completely independent. For example,

some calibration targets require back-lighting, which necessarily requires diﬀerent

lighting than that used for the images of the test piece during the mechanical

test. Also, the lighting and/or exposure may need to be adjusted for diﬀerent

positions/orientations of the calibration target. Additionally, cross-polarized light

may be used to eliminate glare from the target.

Caution 3.9

Any changes in lighting should not disturb the cameras or their mounting. Refer

to Sec. 2.2.4.2 for recommendations on mounting lights.

Caution 3.10

While lighting and exposure may be adjusted, aperture and focus may not be ad-

justed between calibration images of the calibration target and DIC measurement

images of the DIC pattern.

3.2.2.4 Acquire Calibration Images

Acquire calibration images such that there are well-extracted features in the entire

working volume of the optical system, i.e. a volume outlined by the FOV and the

DOF.

Recommendation 3.10

While there are slight variations for diﬀerent software packages, typically the

following positions and orientations of the calibration target are recommended

for stereo-DIC calibration:

1. Rotate about the horizontal image axis.

2. Rotate about the vertical image axis.

3. Plunge towards and away from each camera, along its optical axis.

4. If the calibration target is smaller than the FOV, translate horizontally and

vertically, so that features from the calibration target ﬁll the entire FOV of

each camera.

5. Rotate 90 degrees about the optical axis and repeat the above steps.

6. Perform combinations of the above positions and orientations (i.e. rotate

about the horizontal and vertical axes simultaneously while plunging along

the optical axis).

Manufacturing techniques of calibration targets vary, but some methods may result in a

unidirectional stretch of the pattern on the calibration target. That is, features may have a

slightly diﬀerent spacing along the horizontal axis compared to the vertical axis. While using

targets that have been independently measured is recommended (see Sec. 3.2.2.1), rotating

90 degrees about the optical axis is an additional precaution to help to compensate for any

unidirectional stretch of the pattern on the calibration target.

Tip 3.6

The number of calibration images required or recommended depends on the cal-

ibration target and DIC software, ranging from as few as 8 images up to 50–100

images. Three-dimensional calibration targets (targets that have features on two

diﬀerent planes) may require fewer images than two-dimensional calibration tar-

gets (targets that have features on only a single plane). Consult the user manual

of the software for software-speciﬁc procedures.

Recommendation 3.11

Ideally, a rigid calibration target holder is recommended to ensure that the cal-

ibration target is stationary when images are acquired. The rigid holder can be

The Stereo-DIC Challenge, conducted under the auspices of the Society of Experimental Mechan-

ics, is currently underway. Among other things, the Challenge seeks to explore the eﬀects of the

location and orientation of the calibration target, number of calibration images, etc. Recommenda-

tions for calibration based on this challenge may be included in a future edition of this guide.

mounted on translation and rotation stages or on an adjustable and lockable ball-

in-socket joint, to allow displacement and rotation of the target between images.

In cases where using a rigid calibration holder is not practical, holding cal-

ibration targets by hand is often also acceptable. If the calibration target is

held by hand, hands should be braced against something rigid, and the exposure

time should be limited to approximately 25 ms or less, to reduce motion blur.

Holding calibration targets by hand is not recommended for FOVs smaller than

approximately 50 mm, since even small motions of the target can result in blurry

images.

Caution 3.11

It is more important to have high-quality images (i.e. in focus, good contrast, no

glare, ﬁlling the entire working volume of the optical system, etc.) than to have

a large number of images. Take care to keep the calibration target within the

working volume of the optical system, so that the calibration target remains in

focus, and ensure there is good contrast and no glare on the calibration images.

Recommendation

Some software packages show a live evaluation of the quality of the extraction

of calibration target features, and will only acquire an image if the features are

extracted well. Other software packages extract features only after all calibration

images have been acquired. If using software that follows the second methodology,

a surplus of images can be acquired, allowing for some images and/or features to

be excluded due to poor quality. Be careful, though, not to exclude all images

from a certain region of the working volume of the optical system.

Caution 3.12

For stereo-DIC, the calibration procedure is a minimization process that seeks to

ﬁnd the best set of intrinsic and extrinsic parameters, given a set of extracted

calibration features. Diﬀerent software packages have diﬀerent metrics or “scores”

for the ﬁnal parameter values obtained by the minimization. Often, it is possible

to have a better score with fewer images, or with a smaller volume ﬁlled by

features from the calibration target. This scenario is analogous to obtaining a

higher coeﬃcient of determination (R

value) of a polynomial ﬁt of a set of data

points when fewer points are used. However, if a reduced number of data points

is used in the minimization process, the ﬁnal parameter values may not represent

the optical system with high ﬁdelity.

Recommendation

Take a suﬃcient number of images to have features that ﬁll the working volume

of the optical system, even if the calibration “score” is worse with more images

covering a larger volume, compared to the score with fewer images covering a

smaller volume.

3.2.2.5 Calibrate System

Select an appropriate camera or lens-distortion model,

and calibrate the system with

the DIC software of choice. Refer to the manual of the software for details on the

calibration process.

3.2.2.6 Review Calibration Results

Review the calibration results.

Tip 3.7

This review is dependent on the software utilized; consult the user manual for

speciﬁc suggestions for the DIC software. Some possible aspects of the calibration

results to review (if the software provides access to these aspects) include:

Check the images or features that were rejected and see if there was an

obvious reason for rejection. This is particularly instructive for new users

and/or experienced users working with new hardware setups. It can improve

a user’s ability to produce better quality calibration images in the future

by learning what not to do (i.e. the user can see the eﬀects of poor lighting

or reﬂections, poor ﬁnger or holder placement that blocks key features of

the target, defocused images, etc.).

Verify that the remaining accepted images still ﬁll the working volume of

the optical system. (That is, make sure the rejected images were not all

from the same region of the volume or from the same angle of the calibration

target.)

Verify that the features extracted from the accepted images are correct.

(For example, sometimes the software will extract a feature that is actually

dirt or glare on the calibration target.)

Compare the calibration score from individual images to the score of the

ﬁnal calibration. Also compare the calibration score for a given image from

each camera if using stereo-DIC. Consider removing images manually whose

individual score is signiﬁcantly higher than the overall score, or signiﬁcantly

diﬀerent between the two cameras. Alternatively, consider removing indi-

vidual extracted features so that the individual score of the image is on par

with the overall score.

If possible, save a copy of the individual image calibration scores, in addition

to the overall score and calibration results. This information can be useful

Selection of the camera and lens-distortion model is an advanced topic and is both hardware-

and software-speciﬁc; therefore, no further information is given in the current edition of this guide.

Consult the DIC vendor for more information on selection of the appropriate settings for the software

and hardware.

in diagnosing problems later in the analysis, or problems with one camera

and lens versus another.

Some DIC software packages will alert the user to possible synchronization

errors after the calibration procedure is complete. Synchronization errors

typically only occur if the user is using a hand-held calibration target whose

motion is not completely stopped when images are acquired, or if vibrations

are causing signiﬁcant camera motion when images are acquired. As a di-

agnostic tool, if a synchronization error or camera vibrations are suspected,

one can try to calibrate the intrinsic parameters for each camera individ-

ually. If each camera can be calibrated individually with an acceptable

calibration score, but the extrinsic parameters of the stereo-system cannot

be calibrated, or if the calibration score of the stereo system is unaccept-

able compared to the scores of the individual camera calibrations, then a

synchronization error, or camera vibrations, are likely.

Tip 3.8

The amount of control the user has, and the number of user-deﬁned inputs in the

calibration procedure, varies with diﬀerent DIC software packages. For example,

some software allows the user to select the threshold for extracting features from

the calibration target, or to deﬁne the lens distortion model that is used; other

software is a black box (i.e. closed system) with calibration images as inputs,

and a calibrated camera model and calibration score as outputs. If the DIC

software has any user-deﬁned settings, explore the eﬀect of these settings on the

calibration results.

3.2.2.7 Review Calibration Parameters

Compare the values of the calibration parameters to their corresponding physical values

[29].

Recommendation 3.12

Typical parameters to check include:

Image center: For most camera detector and ﬁxed-focal length lens combi-

nations, the intersection of the optical axis with the detector should be near

the center of the detector array (e.g. if using a 5 MP camera that is 2448

x 2048 pixels, the calibrated image center should be close to (1224,1024)).

This should be true for both cameras of the stereo pair, if the same type

of camera and lens are used. Small variations from the detector center are

to be expected, but non-physical values (e.g. negative values) or extreme

Individual parameters can be completely wrong vis-`a-vis their corresponding physical values, yet

together, the calibration parameters give an accurate triangulation, and accurate displacement results.

Therefore, nonphysical parameters are not necessarily problematic. However, this scenario typically

only arises in complicated DIC measurement setups, and thus is outside of the scope of the current

edition of this guide, which covers only standard laboratory conditions.

values (e.g. near the detector edge) should be investigated and corrected

when possible. When using a zoom lens, however, it is common for the

optical axis to be far from the image center, due to the complexity of the

optics inside of a zoom lens.

Lens focal length: The calibrated focal length of the lens, in physical

units, can be compared to the reported focal length of the physical lens.

Note that the physical lens focal length reported by the lens manufacturer

may be only an approximate value of the actual lens focal length (e.g. a

lens manufacturer may call a certain lens a 50 mm lens, when in reality

the focal length is 47.5 mm). If the calibrated focal length of the lens is

reported in pixels, it can be converted to physical units by multiplying by

the camera detector pixel size in physical units.

Angles:

The reported stereo-angle should be approximately the same as

the physical angle between the two cameras. The other two angles should

be approximately zero if the stereo-plane is perpendicular to the test piece,

as described in Sec. 2.2.2 for standard orientation of the stereo system.

Distance between two cameras:

The reported straight line distance

between the two cameras should be approximately the same as the distance

between the two camera detectors.

This parameter applies only to stereo-DIC and is not applicable to 2D-DIC.

Tip 3.9

Because the physical values are often hard to measure precisely, this review is a

very broad assessment, to make sure the calibration parameters are in the correct

range of values, rather than a very precise comparison between the calibration

parameters and corresponding physical values. Additionally, tracking of these

values for systems that are not changed/adjusted between calibrations can lead

to user experience, and tighter ranges on what is considered acceptable.

Caution 3.13

The calibration score reported by the software can in some cases be misleading.

Refer to Caution 3.12 in Sec. 3.2.2.4 for more details.

3.3 Post-Calibration Routine

The post-calibration routine described in this section has three purposes: to verify

the calibration of the optical system, to acquire images for the noise-ﬂoor analysis

(Sec. 5.4.2), and to perform a ﬁnal review of the DIC system before conducting the

mechanical test with DIC measurements.

3.3.1 Images for Calibration Veriﬁcation and

Noise-Floor Analysis

3.3.1.1 Reset System

Replace the test piece if it was removed to take calibration images. If the stereo system

was moved to take calibration images, replace it to its normal position to view the test

piece, and be sure to lock any moving components on the mounting system.

3.3.1.2 Adjust Lighting

If the lighting and/or exposure were adjusted for the calibration images, readjust light-

ing and/or exposure for the DIC pattern on the test piece.

3.3.1.3 Acquire Static Images

Acquire static images of the test piece.

Recommendation 3.13

Ideally, the static images should be acquired at the same frame rate as that

used for the test, and for the same duration of the test, in order to capture

representative sources of noise or error. For example, high-frequency vibrations

may not be represented if images are acquired at a slow frame rate. Alternatively,

low-frequency errors, such as heat waves or camera drift, may not be represented

if images are acquired over a short duration. Also, for some cameras, camera

noise is a function of frame rate.

Acquiring static images at the same frame rate and for the same duration as

the test, however, doubles the amount of data that must be stored and processed,

which can be non-trivial for many DIC measurements, in which gigabytes of

images and processed data are accumulated. Therefore, the number and timing

of static images is often a compromise between representing the noise sources

present in the DIC measurement, and practical considerations of data size.

One possible strategy to minimize the number of images required for the noise-

ﬂoor, while still representing all noise sources, is to acquire a burst of images at

the desired frame rate at the beginning and the end of the test duration time.

In this way, both high-frequency and low-frequency error sources are captured in

the static images.

3.3.1.4 Review Images

Perform a ﬁnal review of the images from Sec. 3.3.1.3 as described in Sec. 3.1.6.4 and

address any issues that are found.

Caution 3.14

If adjustments are made to the camera(s) or lens(es), the calibration process

will have to be repeated. However, making adjustments to achieve the best-

possible images is usually preferable to producing poor quality or even useless

DIC measurements just to avoid recalibrating the system.

3.3.1.5 Acquire Rigid-Body-Motion Images

Rigidly translate and rotate the test piece and acquire additional images.

Tip 3.10

The applied translations/rotations can either be applied by hand where the exact

applied displacements are unknown, or with translation/rotation stages with mi-

crometers, such that the applied displacements are known within the uncertainty

of the stage.

Recommendation 3.14

At a minimum, translate the test piece within the volume in which it is expected

to move during the test. For a more thorough review of the calibration, acquire

additional images that cover the entire FOV and DOF of each camera.

Recommendation 3.15

For 2D-DIC, capture both in-plane translations and out-of-plane translations and

rotations. The two groups of images — in-plane translations versus out-of-plane

motions — should be kept separate, so that the eﬀects of in-plane versus out-of-

plane motion can be analyzed independently. The in-plane images will be used

to verify adequate correction of lens distortions (Sec. 3.3.2.1) and to calculate the

noise-ﬂoor of QOIs. The out-of-plane motions will be used to estimate the bias

errors caused by out-of-plane motion during the mechanical test (Sec. 5.4.3).

3.3.2 Veriﬁcation of Calibration

Correlate the static and rigid translation images, and verify the calibration results

using the methods described in this section. If the calibration is determined to be

unsatisfactory based on any of the following metrics, improve the calibration before

continuing. Otherwise, accept the calibration, and proceed to a post-calibration review

of the system (Sec. 3.3.3).

Tip 3.11

Improvement may require adjusting software-speciﬁc parameters in the calibra-

tion procedure, taking additional calibration images, or adjusting the optical

system hardware (cameras or lenses).

Caution 3.15

If any adjustments are made to the optical system hardware (camera(s) or lens(es)),

then the previously acquired calibration images must be discarded, an entirely

new set of calibration images must be acquired, and the calibration process must

be redone.

Tip 3.12

The ﬁnal DIC user-deﬁned parameters, i.e. subset size, step size, virtual strain

gauge size, etc. (Sec. 5.2), will not be selected until after the actual mechanical

test has been conducted and a noise-ﬂoor analysis has been completed (Sec. 5.4).

Therefore, at this point, use default settings provided by the software, or expert

judgment and past experience, to select reasonable parameters for the correlation

of the static and translation images for purposes of verifying the calibration and

performing a ﬁnal review of the DIC system.

3.3.2.1 Intrinsic parameters

The primary purpose for verifying intrinsic parameters is to verify that lens distortions

are properly corrected. Correlate the translation images acquired in Sec. 3.3.1 and

remove rigid-body motion. Lens distortions will manifest as an elliptical shape in the

displacement or strain contour plots.

Recommendation 3.16

Evaluation of lens distortions is subjective. Compare the magnitude of the errors

from lens distortions to the total noise-ﬂoor of the displacements and strains

(see Sec. 5.4.2). If errors from lens distortions are signiﬁcant compared to the

noise-ﬂoor, adjust the type and/or magnitude of the distortion correction in the

calibration procedure. If distortions cannot be removed through the correction

process in the calibration procedure, then either a custom correction procedure

will need to be implemented, the choice of optical system (lens and camera)

should be revisited, and/or the image ROI will have to be limited in size and

motion to only the portion of the FOV with an acceptable level of distortion.

Caution 3.16

For 2D-DIC, it is important that the translation images are strictly perpendicular

to the optical axis. Otherwise, false strains due to out-of-plane motion will be

convolved with lens distortions, and the translation images cannot be used to

verify lens distortions alone.

3.3.2.2 Extrinsic parameters

Extrinsic parameters are applicable only for stereo-DIC, and are not applicable for

2D-DIC. The primary metric for verifying the extrinsic parameters is the epipolar er-

ror. Depending on the DIC software, the epipolar error may be called by a diﬀerent

name, such as projection error, three-dimensional residuum, intersection error, or cor-

relation deviation. There are slight diﬀerences in how these metrics are calculated in

diﬀerent software packages, but the basic principle is universal. To verify the extrinsic

parameters, correlate the static images (and translation images if available) acquired

in Sec. 3.3.1 and verify that the epipolar error is acceptable based on the DIC software

documentation.

21,22

Tip 3.13

There is not a single, ﬁxed threshold for the epipolar error that separates “good”

from “bad” calibrations. Rather, there is a direct relationship between epipolar

errors and errors in DIC measurements, with larger epipolar errors resulting in

larger errors in DIC measurements. As a rule of thumb, though, the epipolar

error should typically be on the order of the calibration score; if the epipolar

error is signiﬁcantly larger than the calibration score, the cause of the large error

should be investigated and rectiﬁed.

Tip 3.14

Some DIC software packages only report the average epipolar error over the image

ROI, while others report the epipolar error for each subset. If the DIC software

reports spatially-resolved epipolar error, then the epipolar error can additionally

be used to evaluate the DIC pattern and lighting, similar to using the correlation

residual in the ﬁrst preliminary correlation as described in Sec. 3.1.6.4.

Tip 3.15

Uncorrected lens distortions, in addition to the extrinsic parameters, will also

inﬂuence the epipolar error. However, if the lens distortions are properly corrected

and intrinsic calibration parameters are veriﬁed as described in Sec. 3.3.2.1, then

the epipolar error is primarily related to the extrinsic calibration parameters.

3.3.2.3 Absolute distances

Verify that the DIC measurements are reporting accurate values for absolute distances.

Recommendation 3.17

Some suggested metrics include, but are not limited to:

Fiducial Marks: If ﬁducial marks of a known distance were placed on

the test piece as recommended in Sec. 3.1.5, compare the distance between

the ﬁducial marks calculated by the DIC software in the correlation of

the static images or the translation images to the known distance. This

is only an approximate assessment, since the calculated distance from the

triangulation is known only to within +/- 1 pixel at best, due to manual

See footnote b on page 14 for a note concerning correction of extrinsic parameters of a camera

calibration.

If images of the calibration target were of a diﬀerent image resolution than the static and rigid

translation images of a DIC pattern, then the calibration images must be adjusted for cropping; failure

to do so will result in an inﬂated value of the epipolar error. However, discussion of image cropping

is beyond the scope of the current edition of this guide.

selection of the center of the ﬁducial marks, and there is some uncertainty

in the known distance as well. However, it is a good sanity check to ensure

that the correct target size was entered into (or identiﬁed by) the DIC

calibration software.

Applied Displacements: If the applied displacements are known for the

rigid translation images, compare the DIC results to the applied displace-

ments. This is typically only an approximate assessment, as precision on

DIC results is typically higher than precision of the “known” displacements

if a standard micrometer translation stage is used.

3.3.3 Post-Calibration Review of System

Perform a ﬁnal review of the DIC system. If any aspect of the DIC system is determined

to be unsatisfactory based on the ﬁnal review of the system, adjust the system and

review it again.

Caution 3.17

If any adjustments are made to the optical system hardware (camera(s) or lens(es)),

then the previously acquired calibration images must be discarded, an entirely

new set of calibration images must be acquired, and the calibration process must

be redone.

3.3.3.1 Noise-Floor

Perform an abbreviated noise-ﬂoor analysis and verify that the noise-ﬂoors of the QOIs

are acceptable.

Recommendation 3.18

A full noise-ﬂoor analysis, as described in Sec. 5.4.2, can be time consuming, and

also requires a priori knowledge of the test piece deformation, in order to select

DIC user-deﬁned parameters such as subset size, step size, virtual strain gauge

size, etc. (Sec. 5.2). Therefore, at this point in time, before the mechanical test

has been conducted, an abbreviated noise-ﬂoor analysis is recommended.

Compute the spatial standard deviation of the QOIs from the static images ac-

quired previously (Sec. 3.3.1.3). If the static images were acquired at the desired

frame rate of the actual test, also compute the temporal standard deviation. Ver-

ify that the standard deviations (i.e. the noise-ﬂoor) are acceptable (Sec. 2.1.9).

3.3.3.2 Heat Waves

Look for the presence of heat waves in the displacement contour plots from the static

images. If signiﬁcant heat waves are present, modify the DIC measurement and/or

mechanical test setup to minimize them. See Sec. 2.2.5 for more information.

3.3.3.3 Stability

If a signiﬁcant amount of time passes between calibrating the DIC system and per-

forming the mechanical test with DIC measurements, consider retaking static images

and rechecking the camera calibration. Any increase in the epipolar error or noise-ﬂoor

should be investigated and rectiﬁed.

3.3.3.4 Other Veriﬁcations

In addition to the guidelines outlined here, individual users or laboratories may have ad-

ditional in-house procedures that include details speciﬁc to the mechanical test setups,

equipment, and software that are commonly used in each laboratory. As a matter of

good practice, it is recommended that in-house procedures be documented and that cri-

teria be established to determine if a speciﬁc calibration and/or noise-ﬂoor (Sec. 5.4.2)

is acceptable for the intended purpose of the DIC measurement. This will help prevent

wasting time and resources to complete DIC measurements during a mechanical test,

only to realize after the fact that the images are unsatisfactory.

4 — Execution of the Test

with DIC Measurements

Once all details of the DIC measurement setup and mechanical test setup have been

ﬁnalized, and the cameras have been calibrated, the actual mechanical test can be

conducted, with concurrent imaging for DIC measurements. Before conducting the

test, review all data acquisition systems, such as:

The correct ﬁle name, location, and storage capacity for DIC images has been

set.

The correct test procedure or macro has been selected.

Force signals and other measurement signals from the test frame are set to record

and are synchronized with DIC images.

Triggering of the test frame and/or DIC images is ready.

Caution 4.1

Ensure at least one image is acquired of the test piece prior to any applied

force or displacement.

Lights are turned on, exposure is correct, and frame rate is correct.

As this guide does not cover the mechanical test itself, no further guidelines are

provided here for the actual execution of the test.

5 — Processing of DIC Images

5.1 DIC Software

Once the mechanical test has been performed and DIC images have been acquired, the

images are processed using DIC software. There are both commercial (typically closed-

source) DIC packages as well as independently developed (often open-source) software.

The choice of software depends completely on the user, and the user is directed to the

software manual for speciﬁc details on how to use the software.

As part of the DIC Challenge [14], a set of images to verify DIC software has

been carefully designed and vetted. These images are available at https://sem.org/

dic-challenge/. Users of closed-source DIC software can use these images to explore

the “black box” (closed system) of the DIC software. Additionally, users of indepen-

dently developed DIC codes are strongly encouraged to verify their codes using these

images, and to document the results of the veriﬁcation.

5.2 User-Deﬁned Parameters

There are many user-deﬁned parameters in the DIC analysis procedure that must be

selected. Here, some general comments are made, but detailed training — either on-

the-job training by more experienced DIC practitioners, training by a vendor when

a new DIC system is purchased, or classes taught by subject-matter experts

— is

strongly recommended. Additionally, the user should refer to the manual for the DIC

software of choice for more details that are speciﬁc to the software.

5.2.1 Reference Image

DIC tracks the motion, in a Lagrangian sense, of a set of interrogation points deﬁned

on a reference image. There are three approaches for selecting a reference image:

1. Single reference image: The simplest and preferred approach for selecting a

reference image is to use an image at the beginning of the series, of an undeformed

test piece, prior to the application of any displacement or force. Motion or

displacement of the interrogation points is then tracked over time by correlation

of subsequent images in the series back to the initial reference image.

iDICs oﬀers DIC classes at their annual conference. For more information, see www.idics.org.

Caution 5.1

It is critical that the reference was acquired prior to any applied displace-

ment or force. Otherwise, all DIC measurements correlated with respect to

the reference image will be biased by an unknown amount.

Tip 5.1

One can acquire several (approximately 10) images of the stationary, un-

deformed test piece and average these images together. This averaged ref-

erence image can then be used as an approximately noise-free reference

image.

2. Incremental correlation: In some cases, the DIC pattern may change signiﬁ-

cantly during the course of the test, such that the DIC pattern of the deformed

test piece cannot be correlated back to the initial reference image of the unde-

formed test piece. In this situation, incremental correlation may be used, where

each image is correlated to the previous image, rather than to the same, initial

reference image of the undeformed test piece. Incremental correlation gives in-

cremental displacements between each of the images; total displacements from

the initial reference image of the undeformed test piece are computed by summa-

tion of the incremental displacements. The drawback to incremental correlation,

though, is that errors in the total displacements are also summed, and thus errors

typically increase with increasing number of images in the incremental correlation

sequence.

3. Partitioned correlation: As a compromise between using a single reference

image of the undeformed specimen and incremental correlation, the series of im-

ages may be partitioned into sub-series, and the images in each sub-series are

correlated back to the image at the beginning of that sub-series. For example,

let image 1 be the primary reference image of the undeformed test piece. Then,

images 2-100 may be correlated back to image 1; images 101-200 may be corre-

lated back to image 100; images 201-300 may be correlated back to image 200;

etc. The total displacement of image 300, relative to image 1 of the undeformed

test piece, is then given by the displacement of image 300 relative to image 200

plus the displacement of image 200 relative to image 100 plus the displacement

of image 100 relative to image 1. By updating the reference image periodically

instead of using the previous image, accumulation of errors is reduced.

5.2.2 Pre-Filtering of Images

Subset interpolants often perform better with smooth spatial gradients in image in-

tensity. For this reason, applying a digital low-pass ﬁlter (e.g. a Gaussian ﬁlter) to

the images prior to correlating them, to soften the edges of a particularly sharp DIC

pattern, can be beneﬁcial.

Tip 5.2

Low-pass ﬁlters are also known to mitigate the eﬀects of under-resolved DIC

pattern features (i.e. smaller than 3 pixels) in many cases. Note, though, that

physical anti-aliasing ﬁlters (see Tip 2.11) and digital low-pass ﬁlters are funda-

mentally diﬀerent. The ﬁrst prevents aliasing in the analog realm, so that no

aliased information is encoded in the images. The second attempts to mitigate

the eﬀects of aliasing in the digital realm, after aliased information has been

encoded in the images.

Caution 5.2

Low-pass ﬁltering can also have detrimental eﬀects in some cases (e.g. low-pass

ﬁltering can bias the results). Therefore, DIC practitioners should be judicious

in the use of digital ﬁlters.

More speciﬁc guidelines regarding digital pre-ﬁltering of images may be included in a future

edition of this guide.

Recommendation 5.1

If a physical anti-aliasing ﬁlter was used during image acquisition, digital pre-

ﬁltering is not recommended.

5.2.3 Subset Shape Function

Some DIC software packages ﬁx the subset shape function that is used, while others

allow the user to choose this parameter. When selecting a shape function, there is

a trade-oﬀ between noise ﬁltering and accuracy. Lower order shape functions cancel

more noise, but have less overall accuracy. Higher order shape functions (for example,

quadratic and above) are more accurate, but the standard deviation of the solution will

be higher. There are two outlooks on this trade-oﬀ: One view is to use large subsets

with higher order shape functions. A second view is to use small, closely spaced subsets

with low-order shape functions. An advanced user may explore the diﬀerent options

and combinations, and evaluate which is best for his or her application. See [10, 17, 42]

for more information.

5.2.4 Interpolant

To obtain sub-pixel accuracy of DIC measurements, interpolation of image intensity

between pixels is required. Therefore, the quality of interpolation has a signiﬁcant

inﬂuence on the precision and accuracy of DIC measurements. Most commercial DIC

packages have optimized interpolants, and further reﬁnement is an advanced topic.

For more information, the reader is directed to [18] and [42, Sec. 5.6.1 (Interpolation

Bias)].

5.2.5 Subset Size

Broadly speaking, a subset should be large enough to contain suﬃcient information

such that one subset can be distinguished from all other subsets in the ROI. The rule

of thumb is that the subset should contain a minimum of three DIC pattern features.

If the features are in the optimum 3–5 pixel size range and the feature density is

approximately 50 %, then subsets of approximately 15 × 15 pixels

are required (i.e. a

minimum of three transitions between dark and light pattern features in all directions

is achievable). If the features are larger and/or the feature density is sparse, the subset

size will need to be increased.

Recommendation 5.2

A larger subset size of 21 × 21 pixels

is recommended as a more practical mini-

mum size for typical DIC measurements [17]. This is true particularly if the DIC

pattern size and density is variable and not constant over the entire ROI.

Tip 5.3

Larger subsets typically result in lower displacement noise, but often at the cost

of increased spatial smoothing. Higher-order subset shape functions can be used

to compensate for subset smoothing, though this is an advanced topic.

5.2.6 Step Size

The step size controls the density of points at which DIC data is computed and, to

some extent, inﬂuences the spatial resolution of the measurements. Typically, a step

size of one-third to one-half of the subset size is recommended, so that neighboring

subsets partially overlap, though this value can vary widely depending on speciﬁc ap-

plications. As a general rule, if the overlap is larger than about one-third of the subset

size, then neighboring data points are typically considered as no longer independent,

and decreasing the step size further does not improve the spatial resolution of the

measurements. However, a small step size (in conjunction with a small subset size)

may allow data to be obtained close to the edge or other critical feature of the test

piece, even if the overlap is large and neighboring subsets are not independent. Ad-

ditionally, a small step size may be required to capture the peak position of a QOI

(without interpolation) if it varies quickly across the ROI. (Note, however, that the

peak magnitude of a spatially-varying QOI may still be damped or underestimated due

to the low-pass-ﬁlter eﬀect of DIC, if the spatial resolution is not suﬃcient to capture

the gradients of the QOIs.) If the QOI varies slowly across the ROI, then a large

step size can be used so as to reduce the number of data points, and thus reduce the

computation time. (Even if the QOI varies slowly across the ROI, a maximum step

size equal to the subset size is recommended, to generate quasi-continuous ﬁeld data

without interpolation.) Additionally, the step size also inﬂuences the Virtual Strain

Gauge size (Sec. 5.4.5).

5.2.7 Thresholds

DIC software typically allows the user to select diﬀerent thresholds that are used to

determine the quality and conﬁdence of the displacement results for each subset. The

thresholds available are software-dependent, but two main thresholds include the value

of the matching criterion and the epipolar error. The value of the matching criterion

is a measure of how well each subset was matched between the reference image and a

deformed image (or between the left and right cameras for stereo-DIC). The epipolar

error, which applies only to stereo-DIC, is a measure of how well the correlation results

agree with the stereo calibration. Any displacement results that are above the threshold

values are removed from the reported results. Increasing the threshold values allows

more displacement results to be retained, but at the cost of more uncertainty in the

results.

5.3 Strain Calculations

There are many diﬀerent approaches to calculating strain from displacements, depend-

ing on the speciﬁc DIC software that is used. In each approach, there are diﬀerent

user-deﬁned parameters that can be selected in the software. Refer to the user manual

for explicit details about how strains are computed in the DIC software of choice. In

this section, the virtual strain gauge is deﬁned, and several representative examples of

strain calculation methods are brieﬂy described.

5.3.1 Virtual Strain Gauge (VSG)

One common and key element of all the approaches to strain computation is the virtual

strain gauge (VSG). The VSG, broadly speaking, is the local region of the image that

is used for strain calculation at a speciﬁc location. It is analogous to — though not

directly equal to — the physical area that a foil strain gauge covers. The strain

computed in DIC software is the average or weighted average of the strain within the

VSG.

The exact size of the VSG depends on the method of strain computation used in

the speciﬁc software. Even for a given method of strain computation, the exact size of

the VSG is not well deﬁned. However, there are several key variables that aﬀect the

VSG size, including step size, subset size, subset shape function, strain window, strain

shape function, pre-ﬁltering of the displacements, post-ﬁltering of the strains, and ﬁlter

window. Spatial resolution of strain measurements is closely related to the VSG size,

in addition to other DIC processing parameters. Sec. 5.4.5 provides more information

about the eﬀect of the VSG size on the noise and bias of strain measurements.

5.3.2 Examples of Strain Calculation Methods

Here, four general approaches for strain computation — representative of diﬀerent

approaches implemented in diﬀerent DIC software packages — are brieﬂy described.

The eﬀects of diﬀerent user-deﬁned parameters on the VSG size are highlighted.

5.3.2.1 Subset Shape Function

One approach is to compute the strain directly from the subset shape function and the

deformed subset shape. In this method, the VSG size is approximately equal to the

subset size, giving rise to one of the smallest VSG sizes of all the strain computation

methods. Additionally, no pre-ﬁltering is applied to the displacements. The small VSG

size and the lack of pre-ﬁltering of displacements leads to high spatial resolution of the

strain measurements, but noisy strain results.

After strains are calculated, they may be post-ﬁltered to reduce the noise. A

common type of post-ﬁlter is the mean of a local set of data points, often with a

Gaussian weighting function. The region of the data points that is included in this

ﬁlter is called the ﬁlter window. The VSG size can then be approximated by Eqn. 7.2

(Sec. 7.2).

5.3.2.2 Finite-Element Shape Functions

A second approach closely follows the strain calculations used in ﬁnite-element analysis.

A triangular mesh is deﬁned on the ROI of the reference image, using the displace-

ment data points (provided at the center of the subsets) as the nodes of the mesh.

Using ﬁnite-element shape functions deﬁned over each triangular element, the strain

is computed from the deformed shape of each element. At this point, the VSG size is

small, approximately equal to the size of the triangular elements (which is governed by

the step size) plus the size of the subset. If no pre-ﬁltering of the displacements was

performed, the strain results from this method are usually noisy. Therefore, the strains

are often post-ﬁltered to reduce the noise. A common type of post-ﬁlter is the mean

of a local set of data points, often with a Gaussian weighting function. The region of

the data points that is included in this ﬁlter is called the ﬁlter window. The VSG size

can then be approximated by Eqn. 7.2 (Sec. 7.2).

As a slight variation to the above approach, instead of computing the strain on each

triangular element individually using only the three nodes of the element, the strain

can be computed in a least-squares sense over a larger (i.e. hexagonal) region covering

several triangular elements. This larger region is called the strain window, and the

VSG size be approximated by Eqn. 7.2 (Sec. 7.2). In the least-squares regression, a

weighting function may be applied to the displacement data points contained within

the strain window, for instance with a Gaussian distribution centered at the center of

the strain window and decaying towards the edges of the strain window. By computing

the strain over a larger strain window using more data points, the strain results are

less noisy, and post-ﬁltering of the computed strains may not be necessary.

5.3.2.3 Strain Shape Function

A third approach is to ﬁt a strain shape function to the displacements, which provides

an analytical description of the displacement ﬁeld. The strains are then computed

from the spatial derivatives of this analytical equation. Fitting the displacements to

the strain shape function also serves to ﬁlter the displacements; thus, this ﬁtting process

can be considered as pre-ﬁltering or smoothing the displacements before calculating the

strains. The strain shape function is typically a polynomial or spline ﬁt, and the order

of the strain shape function eﬀects the spatial resolution of the strain measurements.

The local region of the data points that is included in the ﬁt is called the strain

window. Typically, strains are computed at the center of the strain window. A weight-

ing function may be applied to the displacement data points contained within the

strain window, for instance with a Gaussian distribution centered at the center of the

strain window and decaying towards the edges of the strain window. The VSG size is

approximated by Eqn. 7.2 (Sec. 7.2).

5.3.2.4 Spline Fit

A fourth approach to strain calculations is to ﬁt a spline to the entire displacement

ﬁeld, over the entire ROI. This approach is similar to the use of strain shape functions,

except that here, the ﬁt is global rather than local. This spline ﬁt provides an analytical

description of the strains over the entire ROI, which can be evaluated at any point in

the ROI. Thus, strain measurements are not limited to the original DIC data point

locations at the center of the subsets. The VSG size is less clearly deﬁned, since there

is no ﬁlter window or strain window, but it is related to the step size, subset size, and

order of the spline.

5.4 Uncertainty Quantiﬁcation

5.4.1 Overview

There are two types of errors of DIC measurements, i.e. variance errors and bias

errors. Variance errors (also called noise) refer to random errors centered with a mean

about the true value of a QOI. Bias refers to an oﬀset of the mean from the the true

value. The main sources of noise in DIC measurements are camera noise, and matching

errors during the correlation process. Bias can be introduced by smoothing over sharp

spatial gradients of a QOI, uncorrected lens distortions, improper camera calibration

(e.g. if there was relative motion between cameras in a stereo system after calibration

but before the mechanical test), and out-of-plane motion in 2D-DIC measurements, to

name a few sources. Establishing the uncertainty of QOIs — considering both bias and

noise errors — is critical for intelligent assessment and use of DIC results. Without

uncertainty quantiﬁcation, it is impossible to know if a reported QOI value is signiﬁcant

and relevant, or if it is the result of random noise and/or bias, and thus meaningless.

Sec. 5.4.2 and Sec.5.4.3 describe some methods of quantifying noise and bias errors

of DIC measurements. However, bias is often not known, and variance errors computed

from static images acquired prior to the test may not fully represent the variance errors

present during the test. Therefore, the metrics available to a DIC practitioner for

quantifying uncertainty often produce a minimum uncertainty of a QOI, rather than

the true uncertainty. Several options for metrics deﬁning the uncertainty are suggested

here, but other deﬁnitions exist for speciﬁc applications and for diﬀerent QOIs. The

key component, though, is to justify and document (see Sec. 6) some metric and value

for the uncertainty of the QOIs.

5.4.2 Variance Errors

The term “variance error” is used interchangeably with “noise”, and the process of

quantifying the variance errors is often called a noise-ﬂoor analysis. The basic idea of

a noise-ﬂoor analysis is to correlate static images of a DIC pattern that were acquired

under the same conditions as the test images. With no applied force or displacement

on the test piece, all measured QOIs requiring deformation are errors. Any of those

QOIs measured in the actual mechanical test that are smaller than the QOIs measured

from the static images are indistinguishable from the noise.

Recommendation 5.3

A signiﬁcant source of variance errors in DIC measurements is driven by camera

noise. Noise of the camera detector, i.e. ﬂuctuations over time in the gray level

intensity of a pixel observing a ﬁxed object, directly contributes to noise in DIC

results. Therefore, it can be useful to quantify the camera noise independent

of quantifying the noise of the DIC results. This is typically only necessary

when evaluating new hardware for its suitability for DIC (see Sec. 2.2.1). In the

end, the noise-ﬂoor of the QOI is the critical metric, so one may choose to omit

characterizing the noise of the camera itself.

Evaluating the noise-ﬂoor is an iterative process that is typically performed several

times, using sequentially more robust analysis procedures and metrics, during the

design and execution of the DIC measurements. A rudimentary evaluation can be

completed during the preliminary design of the measurements, to aid in the selection

of the camera and lens, choice in patterning technique, etc. (Chapter 2). A second

quick evaluation can be done during the pre-calibration review of the DIC system

(Sec. 3.1.6.4), or the ﬁnal review of the system (Sec. 3.3.3.1), before the mechanical test

is performed. A third evaluation of the noise-ﬂoor is performed iteratively during the

processing of DIC images after the mechanical test is performed, in order to evaluate

the eﬀect of user-deﬁned parameters, and the trade-oﬀ between noise and bias (see

Sec. 5.4.4).

For reporting purposes, the ﬁnal, most thorough evaluation of the noise-ﬂoor must

be done using the same conditions as the mechanical test, both in terms of physical

conditions (e.g. camera and lens selection, lighting, camera temperature, cooling or

mixing fans, test machine powered on) as well as data processing procedures (i.e.

preﬁltering of images, subset size, step size, VSG size, temporal or spatial ﬁltering of

data, etc.). This means that the same user-selected DIC settings that are used for the

analysis of the test piece images during motion/deformation must also be used for the

analysis of the noise-ﬂoor images. Therefore, the ﬁnal noise-ﬂoor analysis is typically

completed after the mechanical test images are analyzed, but using images that were

acquired immediately before the mechanical test.

Using the ﬁnal DIC user-deﬁned parameters, analyze all of the static images that

were acquired after the calibration process (Sec. 3.3.1). Compute any QOIs in the

same manner as was used in the analysis for the mechanical test images. Two dif-

ferent metrics can be used to quantify the variance error of QOIs: a spatial standard

deviation and a temporal standard deviation. To quantify the spatial variation of the

QOI, compute the standard deviation of the QOI for each image. Then average this

spatial standard deviation over time for all the static images. To quantify the temporal

variation, compute the standard deviation of the QOI for each subset over time. Then

average this temporal standard deviation for each subset over all the subsets in the

ROI of the image.

While seemingly similar at ﬁrst glance, these two diﬀerent metrics of error provide

diﬀerent views of the noise-ﬂoor, emphasizing either spatially-varying or temporally-

varying noise. It is recommended to compute both the spatial and the temporal stan-

dard deviation and evaluate if one is signiﬁcantly larger than the other. However,

the spatial and temporal standard deviations are typically similar, and either a single

metric or the average of the two metrics can be selected to quantify the noise-ﬂoor.

Typically, the standard deviation is similar between the horizontal and the vertical

directions. It is recommended to compute the standard deviation for the QOI in each

direction, though, and select either the maximum or the average between the directions.

Given the standard deviation of the QOI, determine the noise-ﬂoor as a function of

the standard deviation, using the experience of a subject matter expert. For example,

some applications may require that all measurements with magnitudes of variation

below three times the standard deviation be considered noise; other applications may

loosen the requirement, so that only measurements with magnitudes of variation below

one standard deviation be considered noise.

5.4.3 Bias Errors

Bias errors are often diﬃcult to quantify, because the true value of a QOI is typically

not known. However, some sources of bias can be evaluated as described below. It

is important to note, however, that these evaluations are necessary but not suﬃcient

to elucidate bias errors. Said another way, some bias errors may be detected through

these evaluations, and if bias errors are detected, they should be reported; however,

even if no bias errors are detected, unknown bias errors may still exist!

One metric of bias error of the QOI is the mean of the QOI from static images.

A mean that changes over time could indicate a bias due to camera drift, heating

of the camera (i.e. the cameras had not yet reached steady state during the camera

warm-up), heat waves, vibrations, etc.

Bias errors due to uncorrected lens distortions can be evaluated from rigid-translation

images, if the rigid-body motion is approximately the same magnitude as the test piece

motion and/or deformation during the actual test. Bias due to uncorrected lens dis-

tortions will manifest as an elliptical shape in the contour plots of strains (and of the

displacements if the mean displacement or known applied displacement is subtracted

from the ﬁeld). This type of bias is typically lower in the center of the FOV and higher

near the edges, due to the mostly radial form of lens distortions.

For 2D-DIC, the bias error due to out-of-plane motion should be evaluated. Using

rigid-body, out-of-plane translation and rotation images, compute the QOI (here, it is

assumed that the QOI is in-plane strain) as a function of applied translation/rotation.

The strain should be zero for rigid-body motion, so any strain measured is a com-

bination of bias and noise. Estimate the amount of out-of-plane translation/rotation

that may have occurred or did occur during the test and report this. Compare the

estimated bias error due to out-of-plane motion to the baseline noise-ﬂoor computed

from static images (Sec. 5.4.2). If the bias error is larger than the variance errors,

consider revising the mechanical test setup to reduce out-of-plane motion.

Bias can also be introduced into the QOI as a result of low-pass ﬁltering, in the

spatial domain, caused by the choices of user-deﬁned parameters. This type of bias

is described in more detail in Sec. 5.4.4. Finally, other factors, such as interpolant,

aliasing, and noise, may cause spatially periodic bias errors that may not be visible in

static images of an unloaded and stationary specimen.

5.4.4 Trade-Off Between Noise and Bias

When selecting user-deﬁned parameters (Sec. 5.2 and Sec. 5.3), there is often a trade-oﬀ

between noise in the measurements and bias due to over-smoothing of the data. Large

subset sizes, low-order subset shape functions, large VSG sizes, pre- or post-ﬁltering

of the data, etc. all reduce noise in the measurements, but at the expense of acting as

low-pass ﬁlters that potentially introduce bias to the measurements. Therefore, when

selecting user-deﬁned parameters, it is important to evaluate their eﬀects on both

noise and bias errors. Often, the ﬁnal selection of parameter values is a compromise

between noise and bias errors. The choice between noisier but unbiased measurements

versus smoother but underestimated measurements is application dependent; expert

judgment is often required to determine which set of parameters produces appropriate

results for a given test. In Sec. 5.4.5, a methodology is presented for evaluating the

trade-oﬀ between noise and bias of strain, since strain is one of the most common QOIs

of DIC measurements. Similar methods, though, can be applied to other QOIs.

The discussion of noise versus bias is also closely tied to the discussion of the spatial

resolution of DIC measurements. Deﬁning the spatial resolution of DIC measurements

is a current topic of interest for iDICs, and iDICs is actively exploring this concept.

For more information on the trade-oﬀ between noise and bias, as well as current eﬀorts

on deﬁning spatial resolution, see [14].

5.4.5 Virtual Strain Gauge Study

Strain is a derived quantity, related to the spatial variation of the displacements. There

are many diﬀerent approaches to calculating strain from displacements, depending on

the speciﬁc DIC software that is used, as described brieﬂy in Sec. 5.3. One common

feature is the requirement that the user select, either directly or indirectly, the size of

the VSG. A virtual strain gauge study is the process used to determine an appropriate

and acceptable VSG size. It also elucidates if the highest spatial strain gradients

that occur in the test piece are captured, and aids in the determination of what the

optimum balance is between capturing spatial strain gradients (i.e. minimizing bias due

to over-smoothing) and improving the strain resolution (i.e. minimizing the variance

errors). The procedure outlined here is based on [38]. Some DIC packages automate

this process, while with others, the user must perform it manually.

For 2D-DIC, bias errors due to poor interpolants or aliasing may be evaluated by translating a ﬂat

test piece out-of-plane towards/away from the camera. This is an advanced topic and is not covered

in this edition of the guide. For stereo-DIC, there are currently no standard procedures for detecting

or evaluating these types of bias errors.

1. Perform an initial DIC analysis on all images of the mechanical test using prede-

termined DIC user-deﬁned parameters, based on vendor defaults or past experi-

ence and expert judgment. Select the image in which the highest strain gradient

is initially calculated.

2. Select the reference image, an image at nominally zero force (acquired after the

reference image), and the image of highest strain gradient determined in the

previous step.

3. Analyze these images with diﬀerent DIC settings, varying the VSG size.

Tip 5.4

The three dominant variables that aﬀect the VSG size are the subset size,

step size, and strain window and/or ﬁlter window. Depending on the strain

calculation method, other variables may also aﬀect the VSG, including but

not limited to the subset shape function and strain shape function.

4. Extract a line cut through the region of highest strain gradient. Plot the strain

along the line for each of the analyses performed in the previous step.

Caution 5.3

Ensure that the line cut does not bridge a crack in the test piece. Computing

strain across a crack is not physically meaningful.

5. As the VSG size decreases, the maximum strain amplitude along the line cut will

typically increase.

If the maximum strain amplitude converges with further decreases of the

VSG, then the actual maximum strain amplitude has been captured. Any

VSG that is larger than the largest VSG that results in the converged actual

maximum strain amplitude will underestimate the actual strain amplitude,

and introduce bias into the measured strain results.

If the maximum strain amplitude never converges, even with the smallest

VSG allowed by the software, then the actual maximum strain amplitude

is unknown. At best, one can report that the actual strain amplitude is

greater than or equal to the maximum measured strain amplitude. That is,

the reported strain is a lower bound on the actual strain amplitude.

Tip 5.5

If the smallest VSG allowed by the software is not suﬃcient, the test

could be repeated with a smaller FOV / higher magniﬁcation. For a

given step size, strain window or ﬁlter window size, etc. — all deﬁned

in terms of pixels in the DIC software — higher magniﬁcation images

would produce a smaller physical VSG size.

6. As the VSG size decreases, the strain noise typically increases, because the

amount of spatial ﬁltering decreases. The noise can be qualitatively assessed

from the line scans used in the previous step. To quantify the noise, compute the

standard deviation of the strain ﬁeld of the zero-force image for each of the DIC

analyses performed.

7. The ﬁnal decision on which VSG size (and other DIC parameters, if those were

varied as well) to use is a matter of expert judgment. If capturing the highest

strain gradient is critical for the DIC analysis, then a small VSG may be the

best choice, even if the noise is large. On the other hand, if there are no high

strain gradients, and/or a smoother strain ﬁeld and/or reduced uncertainty are

more important than knowing the maximum strain at locations of high strain

gradient, then a larger VSG may be the best choice. Alternatively, a combination

of diﬀerent VSG sizes for diﬀerent portions of the test could be appropriate (e.g.

a large VSG size early in the test when the signal-to-noise ratio for strains is

low and strain gradients are small, and a small VSG size later in the test, when

the strain signal-to-noise ratio is higher, and signiﬁcant strain gradients have

developed).

6 — Reporting Requirements

With all the variables that must be selected in a mechanical test with DIC measure-

ments, such as parameters of the physical system (i.e. camera, lens, patterning method,

etc.) and parameters of the data analysis process (i.e. subset size, virtual strain gauge

size, etc.), justiﬁcation and documentation of the choices made is critical. The lists

below present the minimum reporting requirements, as well as suggested and more

detailed reporting recommendations. All documentation of DIC data — both internal

reports and published journal articles — should contain this information.

Tip 6.1

Depending on the application of the DIC data, some of the reporting recommen-

dations may not be necessary, while others not listed here may be important.

The key, though, is to document all relevant information!

6.1 DIC Hardware Parameters

6.1.1 Required

Camera Manufacturer and Model, and Image Resolution

Lens Manufacturer and Model, and Focal Length

Note 1: If lens has a variable focal length, report both range and focal length

used.

FOV

Image Scale

Note 1: In stereo-DIC, where the cameras are at an angle to the test piece,

the image scale is not constant across the FOV, and can be diﬀerent in

the two cameras. Therefore, the image scale of the ROI of the image

should be reported, either as the average for the two cameras, if the

scale is nearly the same, or for each camera individually, if the scale is

signiﬁcantly diﬀerent in the two cameras.

Stereo-Angle

Note 1: Applicable for stereo-DIC; not applicable for 2D-DIC.

SOD

Image Acquisition Rate

Patterning Technique

Approximate Pattern Feature Size

Note 1: Specify method used to determine feature size. Note that both light

(white) and dark (black) regions are considered features.

6.1.2 Recommended

Aperture

Image Noise

6.2 DIC Analysis Parameters

6.2.1 Required

DIC Software Package Name and Manufacturer

Note 1: If an independently developed (non-commercial) DIC code is used, it

is strongly recommended to verify the code using images from the DIC

Challenge [14] (https://sem.org/dic-challenge/). Any subsequent

documentation of DIC measurements that use the code should refer to

this veriﬁcation.

Image Filtering, if applied

Subset Size

Note 1: Preferably, report both in terms of pixels and in terms of physical

units (e.g. millimeter) by scaling based on the image scale.

Step Size

Note 1: Preferably, report both in terms of pixels and in terms of physical

units (e.g. millimeter) by scaling based on the image scale.

Subset Shape Function (e.g. aﬃne, quadratic)

Data Processing and Filtering for QOIs

Recommendation 6.1

Strain is one of the most common QOIs. Typical parameters to report

include:

– Pre-ﬁltering of displacements (spatial and/or temporal), if applied

– Strain formulation (i.e. Lagrange, engineering, logarithmic etc.)

– Strain window

– Virtual strain gauge size

Note 1: Preferably, report both in terms of pixels and in terms of

physical units (e.g. millimeter) by scaling based on the image

scale.

– Post-ﬁltering of strains (spatial and/or temporal), if applied

Noise-Floor and Bias of QOIs

Note 1: For 2D-DIC, bias caused by out-of-plane motion should be reported.

6.2.2 Recommended

DIC Software Package Version Number

Calibration parameters, such as the following list. Note that models for cali-

bration parameters are software-speciﬁc. The parameters listed here represent

one particular model; relevant parameters for the selected model used should be

reported.

– Model number and serial number of calibration target used. (This informa-

tion is useful for traceability and to elucidate any errors in measurements

that may be associated with a speciﬁc, physical calibration target.)

– Image Center

Note 1: Report for both cameras if using stereo-DIC.

– Focal Length

Note 1: Report for both cameras if using stereo-DIC.

– Lens Distortion Correction Model and Parameters

– Stereo-Angle

Note 1: Applicable for stereo-DIC; not applicable for 2D-DIC.

– Distance between Cameras

Note 1: Applicable for stereo-DIC; not applicable for 2D-DIC.

– Calibration Quality Metric(s)

Interpolant

Matching Criterion

7 — Glossary and Acronyms

7.1 Acronyms

DIC: Digital Image Correlation

DOF: Depth-of-Field

FOV: Field-of-View

iDICs: International Digital Image Correlation Society

QOI: Quantity-of-Interest

ROI: Region-of-Interest

SOD: Stand-Oﬀ Distance

VSG: Virtual Strain Gauge

7.2 Glossary

Calibration Score: The residual of the bundle adjustment optimization process used

to calibrate a DIC system.

Digital Image Correlation: Within the scope of this guide, Digital Image Correla-

tion (DIC) is an optically-based technique used to measure the evolving full-ﬁeld

2D or 3D displacements on the surface of a test piece, throughout a mechanical

test of a material or structure.

Note 1: 2D-DIC refers to the measurement of displacements in only two direc-

tions on the surface of the test piece, where one camera is oriented perpen-

dicularly to a planar test piece.

Note 2: Stereo-DIC refers to the measurement of shape and displacements in

three directions on the surface of the test piece, by using two (or more)

cameras oriented at diﬀerent angles. Stereo-DIC is sometimes called 3D-

DIC, but should not be confused with volumetric-DIC, which provides shape

and displacement measurements throughout the volume of the test piece.

Data Filtering: Any further post-processing of the results to spatially or temporally

ﬁlter the DIC results (could include a Gaussian ﬁlter, median ﬁlter, etc.)

Data Point: A point at which DIC results (displacements, strains, etc.) are reported.

Data points are typically reported at the center of subsets in local DIC.

Dynamic Range, Detector [counts or gray levels]: Number of bits of the analog

to digital converter of a camera detector (e.g. 8-bit).

Dynamic Range, Image [counts or gray levels]: Range of gray levels contained

in the image data. This can be graphically viewed in the image histogram. The

image dynamic range is less than or equal to the detector dynamic range.

Epipolar Error [pixel]: The distance between the location of a data point, as de-

termined by cross-correlation of a pair of images from the two cameras of a

stereo-DIC system, and the epipolar line.

Note 1: Depending on the DIC software, the epipolar error may also be called

projection error, three-dimensional residuum, intersection error, or correla-

tion deviation.

Note 2: The epipolar line is determined by the extrinsic parameters of the stereo-

camera calibration (i.e. stereo-angle, distance between two cameras). For

more information on epipolar geometry, refer to [42, Sec. 4.2 (Three-Dimensional

Computer Vision)].

Field-of-View (FOV) [mm × mm]: The region of space projected through a lens

system onto a camera detector.

Gray Level [counts]: The image intensity recorded by the image acquisition system,

expressed as the number of counts of the digitizer.

Note 1: This value is proportional to the measured light intensity, but typically

has no absolute calibrated relationship to the measured intensity. For DIC,

this lack of calibration is acceptable, because the image is used for tracking

the object motion, rather than measuring the light intensity at points on

the object.

Note 2: Usually the number of counts is relative to the number of bits (quanti-

zation level) in the imaging analog-to-digital converter.

Image Data: Recorded “images” of a test piece containing encoded information re-

lated to the displacement ﬁeld including displacement gradients, nearly always a

2D or 3D numerical array of “intensity” or gray level data that will be used for

correlation.

Image Filtering: Any type of image data processing done to modify the gray level

values of the pixels, most often a smoothing operation.

Note 1: Analog Image Filtering refers to ﬁltering that is done in an analog fashion

by modifying the physical optical system, e.g. with a blur ﬁlter assembled

on the camera detector or by defocusing the lens.

Note 2: Digital Image Filtering refers to ﬁltering that is done in a digital fashion

as a post-processing step after the image has been acquired, e.g. a Gaussian

ﬁlter.

Image Noise [counts or gray levels or percent of dynamic range]: Pixel-wise ac-

quisition noise of the imaging system. This often varies depending on pixel in-

tensity, camera temperature and optical intensity.

Image Scale [pixel/mm]: Number of optical elements (pixels) used to record an

image of a region of physical length. The image scale can be used to convert

from the image pixel size to physical units (e.g. meter).

Note 1: The image scale varies with position in an image. In 2D-DIC, with a

single camera perpendicular to the test piece, the variation tends to be small,

since the variation is the result of lens distortions. In stereo-DIC, where the

cameras are angled with respect to the surface of interest, the variation in

image scale is much larger. This is the result of a combination of the lens

distortions and the perspective eﬀect (which is reversed in the left and right

images). For stereo-DIC systems, the average image scale of the ROI shall

be reported.

Interpolant: Interpolation function used to calculate the subpixel changes within the

subset shape function transformation subject to the matching criteria during the

correlation calculation.

Matching Criterion: Mathematical formulation used to calculate the quality metric

of the calculated displacement ﬁeld based on the underlying image data. Also

commonly referred to as “correlation criterion.”

Note 1: Common matching criteria include, but are not limited to, sum of square

diﬀerences (SSD), normalized sum of square diﬀerences (NSSD), zero-normalized

sum of square diﬀerences (ZNSSD) and cross-correlation (CC).

Noise-Floor: [See Resolution of a Quantity-of-Interest.]

Pattern Feature Size [pixel]: Characteristic length (e.g. diameter) of DIC pattern

features in the image data, reported in terms of pixels.

Note 1: For DIC patterns that consist of primarily circular features (i.e. speck-

les), the pattern feature size is sometimes referred to as the “speckle size.”

Note 2: If a range of feature sizes exist in the image, the mean size and an

indication of the distribution of sizes (e.g. minimum and maximum, or

standard deviation) should be reported.

Note 3: Physical size of the features can be calculated by dividing by the image

scale.

Note 4: The spatial frequency of the pattern can be determined as the inverse of

the pattern feature size (e.g. 1/(pattern feature size)).

Pixel: Region over which the image data is averaged and quantized. There is a re-

sulting gray level or number of counts at each pixel relative to some underlying

input, usually optical intensity.

Quantity-of-Interest (QOI): An attribute or property of a test piece that may be

distinguished qualitatively and determined quantitatively [8], which a person

seeks to characterize by performing a particular test.

Note 1: QOIs may be both direct measurements or derived quantities. With

respect to DIC, common QOIs are shape, curvature, displacement, velocity,

acceleration, strain, strain-rate, etc.

Quantization Level [bits]: Number of bits used to record the gray level at each

pixel. This may be light intensity for optical images, X-ray density for computed

tomography, or any other information encoded as image contrast (image data).

(A height map in an atomic force microscope is an example of a diﬀerent type of

“image data”.)

Region-of-Interest (ROI) of the Test Piece [mm × mm]: The portion of sur-

face of the test piece that is used for analysis.

Note 1: The term “area-of-interest” is sometimes used interchangeably with the

term “region-of-interest.”

Note 2: The region may be of any arbitrary shape, and may change shape in

consecutive images.

Note 3: The term “region-of-interest” can refer to either a portion of the test

piece or the corresponding portion of an image, and context typically is

suﬃcient to distinguish between the two demarcations.

Region-of-Interest (ROI) of the Image [pixel × pixel]: The portion of the im-

age corresponding to the region-of-interest of the test piece.

Note 1: The term “area-of-interest” is sometimes used interchangeably with the

term “region-of-interest.”

Note 2: All QOIs are measured or derived using the image data that comes from

the ROI of the image.

Note 3: The term “region-of-interest” can refer to either a portion of the test

piece or the corresponding portion of an image, and context typically is

suﬃcient to distinguish between the two demarcations.

Resolution, Image [pixel × pixel]: Total number of pixels contained in an image,

typically reported as the width by height of the detector array in pixels.

Note 1: Image resolution should not be confused with optical resolution or spatial

resolution.

Resolution, Optical [line pair / mm]: The ability of an imaging system to resolve

detail in the object being imaged.

Note 1: Optical resolution is typically measured from images of a resolution tar-

get.

Resolution, Spatial [pixel]: The minimum distance between two localized features

that can be independently resolved.

Note 1: This deﬁnition might be counter intuitive, in that a smaller resolution

value is desireable, whereas a larger resolution value is generally less de-

sirable. These trends are opposite those of image resolution and optical

resolution.

Note 2: For the current edition of this guide, the concept of spatial resolution

is deﬁned as above; however, a uniﬁed method to determine the spatial

resolution of DIC measurements is a current topic of interest for iDICs, and

iDICs is actively exploring this concept in more detail.

Resolution Target: An object with features of speciﬁed width and/or spacing, used

to determine the optical resolution of an imaging system.

Note 1: Two common resolution targets are the 1951 USAF resolution target or

the Siemens star, which can be purchased from major optics companies. See

https://en.wikipedia.org/wiki/1951_USAF_resolution_test_chart and

https://en.wikipedia.org/wiki/Siemens_star for more information.

Resolution of a Quantity-of-Interest: The threshold value of a QOI below which

measurements are indistinguishable from noise, and above which measurements

are signiﬁcant.

Note 1: The phrase “Resolution of a QOI” is used interchangeably with the

phrase “noise-ﬂoor” in this guide.

Note 2: The noise-ﬂoor is typically deﬁned as a multiple of the standard deviation

(either spatial or temporal) of the QOI computed under conditions in which

the QOI should be zero.

Note 3: The noise-ﬂoor reﬂects only the random variance error of the QOI, and

does not reﬂect any systematic bias errors that may be present in the QOI.

See Sec. 5.4 for more information on variance versus bias errors.

Shape Function, Strain: Analytic equation that is ﬁt, in a least-squares sense, to

the displacement data within the strain window. Strains are computed from the

derivatives of this equation.

Note 1: The strain shape function should not be confused with the subset shape

function.

Note 2: Not all methods of computing strain invoke a strain shape function.

Shape Function, Subset: Equation used to describe the displacement ﬁeld within a

subset.

Note 1: Aﬃne (linear) is the most common subset shape function, but higher

ordered implementations are also used.

Note 2: The subset shape function should not be confused with the strain shape

function.

Stand-Oﬀ Distance [m]: The distance between the aperture of the lens and the test

specimen.

Stereo-Angle [degree]: In a stereo-DIC system, the included angle between the op-

tical axis of each of the two camera systems (i.e. camera and lens).

Stereo-Plane: In a stereo-DIC system, the plane formed by the optical axes of the

two camera systems (i.e. camera and lens).

Step Size, L

step

[pixel]: The spacing of pixel grid points at which the subset dis-

placements are calculated. That is, there will be a displacement solution at

every step in the ROI.

Note 1: The step size is also sometimes reported as overlap. For example, 50%

overlap means a step size of half the subset size.

Subset: Portion of the image that is used to calculate one 3D coordinate value, or

one displacement value.

Note 1: Center point displacement is commonly reported, although other param-

eters may be available via the subset shape function.

Subset Size, L

subset

[pixel]: Length of the subset in the reference image.

Note 1: Subsets are typically square or circular (in the reference image), and

thus a single length is suﬃcient to deﬁne the subset size. Some software,

however, permits rectangular subsets; in this case, dimensions of both sides

of the rectangle should be given to deﬁne the subset size.

Weighting Function: Mathematical device used to give some elements more inﬂu-

ence on a result than other elements, based on the spatial location of the elements.

Note 1: Common weighting functions are square or uniform (which weights all

elements equally) or Gaussian (which weights elements closer to the center

point of interest more heavily than elements farther from the center point

of interest).

Note 2: A subset weighting function is used to weight the intensities of the pixels

contained within the subset when performing subset matching.

Note 3: A strain weighting function is used to weight the displacement data

points within the strain window when computing strain.

Note 4: A ﬁlter weighting function is used to weight the data within a ﬁlter

window when applying a spatial data ﬁlter.

Window, Filter: Local region of the ROI of the image, containing a ﬁnite number

of data points, that is used for local spatial ﬁlters of DIC data.

Note 1: See Window Size for information about the ﬁlter window size.

Window, Strain: Local region of the ROI of the image, containing a ﬁnite number

of data points, that is used to calculate strain.

Note 1: Not all methods of computing strain invoke a strain window.

Note 2: See Window Size for information about the strain window size.

Window Size, L

window

[data point]: Characteristic length of a local region of data

points (e.g. a ﬁlter window or a strain window).

Note 1: Strain and ﬁlter windows are typically square, circular, or hexagonal, and

the size of the window is given by the characteristic length of the window (i.e.

one side of the square, the diameter of the circle, or the eﬀective diameter

of the hexagon). The window size is speciﬁed in terms of the number of

data points that span the characteristic length of the window. Windows are

typically symmetric and centered at a data point; thus, window sizes are

typically odd integers.

Note 2: The window size in terms of pixels, L

∗

window

, is given by Eqn. 7.1, where

window

is the window size in terms of data points, and L

step

is the step

size.

∗

window

= (L

window

− 1) L

step

(7.1)

Note 3: To determine the window size in terms of physical units, the window size

in terms of pixels must be divided by the average image scale.

Virtual Strain Gauge (VSG): The local region of the image that aﬀects the strain

value at a speciﬁc location.

Note 1: The VSG is analogous to — but not exactly equal to — the physical

area that a physical strain gauge would cover.

Virtual Strain Gauge Size, L

V SG

[pixel]: Characteristic length of the virtual strain

gauge.

Note 1: Virtual strain gauges are typically square, circular, or hexagonal, and

the size of the VSG is given by the characteristic length of the VSG (i.e.

one side of the square, the diameter of the circle, or the eﬀective diameter

of the hexagon). The VSG size is speciﬁed in terms of the number of pixels

that span the characteristic length of the VSG.

Note 2: The size of the VSG depends on the strain calculation method and user-

deﬁned parameters such as step size, subset size, strain window, ﬁlter win-

dow, strain shape function, weighting functions, and subset shape function.

An estimate for the size of the VSG, if L

window

> 0, is given by Eqn. 7.2,

where L

window

is the window size (of either the strain window or of the ﬁlter

window), L

step

is the step size, and L

subset

is the subset size.

V SG

= (L

window

− 1) L

step

+ L

subset

(7.2)

Note 3: To determine the VSG size in terms of physical units, the VSG size must

be divided by the average image scale.

References

[1] Blaysat, B., Gr´ediac, M., Sur, F. (2016) On the Propagation of Camera Sensor

Noise to Displacement Maps obtained by DIC — An Experimental Study. Exp

Mech, 56(6):919–944. https://doi.org/10.1007/s11340-016-0130-9.

[2] Bossuyt, S. (2013) Optimized Patterns for Digital Image Correlation. Proceedings

of the 2012 Annual Conference on Experimental and Applied Mechanics. Vol 3:

Imaging methods for Novel Materials and Challenging Applications, pg 239–248.

Springer, New York. https://doi.org/10.1007/978-1-4614-4235-6_34.

[3] Byrne, E., Simonsen, M. (2016) Lens Selection and Stereo Angle. Accessed

Sept. 2018. http://www.correlatedsolutions.com/support/index.php?

/Knowledgebase/Article/View/50/1/lens-selection-and-stereo-angle

[4] Crammond, G., Boyd, S.W., Dulieu-Barton, J.M. (2013) Speckle Pattern Qual-

ity Assessment for Digital Image Correlation. Optics and Lasers in Engineering

51(12):1368–1378. https://doi.org/10.1016/j.optlaseng.2013.03.014.

[5] Dong, Y.L., Pan, B. (2017) A Review of Speckle Pattern Fabrication and As-

sessment for Digital Image Correlation. Exp Mech. 57(8):1161–1181. https:

//doi.org/10.1007/s11340-017-0283-1.

[6] Gr´ediac, M., Hild, F., Pineau, A. (2013) Full-Field Measurements and Identiﬁ-

cation in Solid Mechanics. Wiley Online Library: New York. https://doi.org/

10.1002/9781118578469.

[7] Iadicola, M.A. (2016) Uncertainties of Digital Image Correlation due to Pattern

Degradation at Large Strain. In: Jin, H., Yoshida, S., Lamberti, L., Lin, M.T.

(eds) Advancement of Optical Methods in Experimental Mechanics. Vol. 3. Con-

ference Proceedings of the Society for Experimental Mechanics. Springer, Cham.

https://doi.org/10.1007/978-3-319-22446-6_31.

[8] JCGM Member Organizations (2012) International Vocabulary of Metrology —

Basic and General Concepts and Associated Terms (VIM). Edition 3. Vol 200.

BIPM. https://www.bipm.org/en/publications/guides/vim.html.

[9] Jones, E.M.C., Reu, P.L. (2017) Distortion of Digital Image Correlation (DIC)

Displacements and Strains from Heat Waves. Exp Mech. https://doi.org/10.

1007/s11340-017-0354-3.

[10] Lava, P., Cooreman, S., Coppieters, S., De Strycker, M., Debruyne, D. (2009)

Assessment of Measuring Errors in DIC using Deformation Fields Generated by

Plastic FEA. Optics and Lasers in Engineering 47(7-8):747–753. https://doi.

org/10.1016/j.optlaseng.2009.03.007.

[11] Lecompte, D., Smits, A., Bossuyt, S., Sol, H., Vantomme, J., Hemelrijck, D.V.,

Habraken, A. (2006) Quality Assessment of Speckle Patterns for Digital Image

Correlation. Optics and Lasers in Engineering, 44(11): 1132–1145. https://

doi.org/10.1016/j.optlaseng.2005.10.004.

[12] LePage, W.S., Daly, S.H., Shaw, J.A. (2016) Cross Polarization for Improved

Digital Image Correlation. Exp Mech, 56(6):1–17. https://doi.org/10.1007/

s11340-016-0129-2.

[13] Pan, B., Yu, L., Wu, D. (2014) High-Accuracy 2D Digital Image Correlation

Measurements using Low-Cost Imaging Lenses: Implementation of a Generalized

Compensation Method. Meas Sci Technol, 25(2):025001. https://doi.org/10.

1088/0957-0233/25/2/025001.

[14] Reu, P.L., Toussaint, E., Jones, E.M.C., Bruck, H.A., Iadicola, M., Balcaen,

R., Turner, D.Z., Siebert, T., Lava, P., Simonsen, M. (2017) DIC Challenge:

Developing Images and Guidelines for Evaluating Accuracy and Resolution of

2D Analyses. Exp Mech. https://doi.org/10.1007/s11340-017-0349-0.

[15] Reu, P.L. (2013) A Study of the Inﬂuence of Calibration Uncertainty on the

Global Uncertainty for Digital Image Correlation Using a Monte Carlo Approach.

Exp Mech, 53:1661-1680. https://doi.org/10.1007/s11340-013-9746-1.

[16] Reu, P.L. (2012) Introduction to Digital Image Correlation: Best Practices and

Applications. In “The Art and Application of DIC” article series. Exp Techniques,

36(1):3–4. https://doi.org/10.1111/j.1747-1567.2011.00798.x.

[17] Ibid, (2012) Hidden Components of DIC: Calibration and Shape Function —

Part 1. 36(2):3–5. https://doi.org/10.1111/j.1747-1567.2012.00821.x.

[18] Ibid, (2012) Hidden Components of 3D-DIC: Interpolation and Matching — Part

2. 36(3):3–4. https://doi.org/10.1111/j.1747-1567.2012.00838.x.

[19] Ibid, (2012) Hidden Components of 3D-DIC: Triangulation and Post-Processing

— Part 3. 36(4):3–5. https://doi.org/10.1111/j.1747-1567.2012.00853.x.

[20] Ibid, (2012) Stereo-Rig Design: Creating the Stereo-Rig Layout — Part 1.

36(5):3–4. https://doi.org/10.1111/j.1747-1567.2012.00871.x.

[21] Ibid, (2012) Stereo-Rig Design: Camera Selection — Part 2. 36(6):3–4. https:

//doi.org/10.1111/j.1747-1567.2012.00872.x.

[22] Ibid, (2013) Stereo-Rig Design: Lens Selection — Part 3. 37(1):1–3. https:

//doi.org/10.1111/ext.12000.

[23] Ibid, (2013) Stereo-Rig Design: Stereo-Angle Selection — Part 4. 37(2):1–2.

https://doi.org/10.1111/ext.12006.

[24] Ibid, (2013) Stereo-Rig Design: Lighting — Part 5. 37(3):1–2. https://doi.

org/10.1111/ext.12020.

[25] Ibid, (2013) Calibration: Pre-Calibration Routines. 37(4):1–2. https://doi.

org/10.1111/ext.12026.

[26] Ibid, (2013) Calibration: 2D Calibration. 37(5):1–2. https://doi.org/10.

1111/ext.12027.

[27] Ibid, (2013) Calibration: A Good Calibration Image. 37(6):1–3. https://doi.

org/10.1111/ext.12059.

[28] Ibid, (2014) Calibration: Stereo Calibration. 38(1):1–2. https://doi.org/10.

1111/ext.12048.

[29] Ibid, (2014) Calibration: Sanity Checks. 38(2):1–2. https://doi.org/10.1111/

ext.12077.

[30] Ibid, (2014) Calibration: Care and Feeding of a Stereo-Rig. 38(3):1–2. https:

//doi.org/10.1111/ext.12083.

[31] Ibid, (2014) Speckles and their Relationship to the Digital Camera. 38(4):1–2.

https://doi.org/10.1111/ext.12105.

[32] Ibid, (2014) All about Speckles: Aliasing. 38(5):1–3. https://doi.org/10.

1111/ext.12111.

[33] Ibid, (2014) All about Speckles: Speckle Size Measurement. 38(6):1–3. https:

//doi.org/10.1111/ext.12110.

[34] Ibid, (2015) All about Speckles: Contrast. 39(1):1–2. https://doi.org/10.

1111/ext.12126.

[35] Ibid, (2015) All about Speckles: Edge Sharpness. 39(2):1–2. https://doi.org/

10.1111/ext.12139.

[36] Ibid, (2015) All about Speckles: Speckle Density. 39(3):1–2. https://doi.org/

10.1111/ext.12161.

[37] Ibid, (2015) Points on Paint. 39(4):1–2. https://doi.org/10.1111/ext.12147.

[38] Ibid, (2015) Virtual Strain Gauge Size Study. 39(5):1–3. https://doi.org/10.

1111/ext.12172.

[39] Ibid, (2015) DIC: A Revolution in Experimental Mechanics. 39(6):1–3. https:

//doi.org/10.1111/ext.12173.

[40] Reu, P.L. (2018) Using Anti-Aliasing Camera Filters for DIC: Does it Make a

Diﬀerence?. In: Lamberti, L., Lin, M.T., Furlong, C., Sciammarella, C. (eds)

Advancement of Optical Methods in Experimental Mechanics. Vol. 3. Conference

Proceedings of the Society for Experimental Mechanics. Springer, Cham. https:

//doi.org/10.1007/978-3-319-63028-1_14.

[41] Reu, P.L., Sweatt, W., Miller, T., Fleming, D. (2015) Camera System Resolution

and its Inﬂuence on Digital Image Correlation. Exp Mech, 55:9–25. https://doi.

org/10.1007/s11340-014-9886-y.

[42] Sutton, M.A., Orteu, J.J., Schreier, H. (2009) Image Correlation for Shape, Mo-

tion and Deformation Measurements. Springer US. https://doi.org/10.1007/

978-0-387-78747-3.

[43] Sutton, M.A., Yan, J.H., Tiwari, V., Schreier, H.W., Orteu, J.J. (2008) The

Eﬀect of Out-of-Plane Motion on 2D and 3D Digital Image Correlation Measure-

ments. Optics and Lasers in Engineering 46(10)746–757. https://doi.org/10.

1016/j.optlaseng.2008.05.005.

[44] Wittevrongel, L., Badaloni, M., Balcaen, R., Lava, P., Debruyne, D. (2015)

Evaluation of Methodologies for Compensation of Out-of-Plane Motions in 2D

Digital Image Correlation Setup. Strain, 51(5):357–369. https://doi.org/10.

1111/str.12146.

[45] Zappa, E., Mazzoleni, P., Matinmanesh, A. (2014) Uncertainty Assessment of

Digital Image Correlation Method in Dynamic Applications. Opt Laser Eng,

56:140–151. https://doi.org/10.1016/j.optlaseng.2013.12.016.

A — Checklist and Flow Chart

for DIC Measurements

and Analysis

This appendix presents a checklist and ﬂow chart of the main points to consider when

designing, executing, and analyzing DIC measurements performed during mechanical

testing of a planar test piece. Each of the steps listed in the checklist are expounded

upon in the main body of this guide, and the ﬂow chart (Fig. A.1) refers in parentheses

to speciﬁc sections of the guide.

1. Design of DIC Measurements (2)

(a) Measurement Requirements

 QOIs (2.1.1)

 ROI (2.1.2)

 FOV (2.1.3)

 Position Envelope for Hardware (2.1.4)

 2D-DIC vs Stereo-DIC (2.1.5)

 Stereo-Angle (2.1.6)

 DOF (2.1.7)

 Spatial Gradients (2.1.8)

 Noise-Floor (2.1.9)

 Frame Rate (2.1.10)

 Exposure Time (2.1.11)

 Synchronization and Triggering (2.1.12)

(b) Equipment Selection

 Camera and Lens (2.2.1)

 Mounting Equipment (2.2.2)

 Aperture (2.2.3)

 Lighting and Exposure (2.2.4)

 DIC pattern (2.3)

 Test DIC pattern technique on extra test piece(s).

 Evaluate DIC pattern behavior throughout test.

 Evaluate lighting/contrast throughout test.

 Evaluate data synchronization and triggering.

2. Preparation for the Measurements (3)

(a) Pre-Calibration Routine (3.1)

 Review test procedure (3.1.1).

 Check cleanliness of camera detector, lens, and calibration target (3.1.2).

 Warm up cameras (3.1.3).

 Synchronize cameras to each other and to other data acquisition (3.1.4).

 Apply DIC pattern (3.1.5).

(b) Pre-Calibration Review of System (3.1.6)

 Position test piece in load frame (3.1.6.1).

 Position cameras for desired FOV and image ROI (3.1.6.1).

 Verify FOV, focus, DOF (3.1.6.2).

 Lock all moving parts of cameras, lenses, and mounting system (3.1.6.3).

 Adjust orientation of polarization ﬁlters if using cross-polarized light

(3.1.6.3).

 Review static images (3.1.6.4), looking for:

Glare

DIC pattern that is too coarse or too ﬁne

Defects in applied DIC pattern

Out-of-focus regions of the image

Poor contrast

Non-uniform lighting

Overexposed or underexposed regions

Dirt, smears, foreign object on lens or camera detector

Vibrations or other camera motion

 Adjust DIC system until high-quality images are obtained.

 Select calibration target of appropriate size. (3.2.2.1).

 Create a clear working space in which to perform calibration (3.2.2.2).

 Lock all moving parts of cameras, lenses, and mounting system (3.2.2.2).

 Adjust lighting/exposure (3.2.2.3).

 Ensure there is uniform contrast and no glare as the calibration target

is rotated, tilted, and translated (3.2.2.3).

 Acquire calibration images that have well-extracted features in the en-

tire working volume of the optical system (3.2.2.4).

 Calibrate the system (3.2.2.5).

 Review calibration results (3.2.2.6).

 Review calibration parameters (3.2.2.7).

(d) Post-Calibration Routine (3.3)

 Reset system: Position test piece in test frame (if removed for calibra-

tion) or reposition stereo-camera system (if moved for calibration) and

lock any moving parts (3.3.1.1).

 Adjust lighting/exposure (3.3.1.2).

 Acquire static images (3.3.1.3).

 Review static images (3.3.1.4 and 3.1.6.4), looking for:

Glare

DIC pattern that is too coarse or too ﬁne

Defects in applied DIC pattern

Out-of-focus regions of the image

Poor contrast

Non-uniform lighting

Overexposed or underexposed regions

Dirt, smears, foreign object on lens or camera detector

Vibrations or other camera motion

 Acquire rigid-body-motion images of test piece for noise-ﬂoor analysis

(3.3.1.5).

 Verify calibration (3.3.2).

Intrinsic parameters (3.3.2.1)

Extrinsic parameters (3.3.2.2)

Absolute distances (3.3.2.3)

 Perform abbreviated noise-ﬂoor analysis and ensure the noise-ﬂoor is

acceptable (3.3.3.1).

 Look for heat waves (3.3.3.2), system stability (3.3.3.3), and any other

lab-speciﬁc system veriﬁcations (3.3.3.4).

3. Execution of the Test with DIC Measurements (4)

 Verify correct ﬁle name, location, and storage capacity for DIC images.

 Verify that the correct test procedure or macro has been selected.

 Verify force and other measurements of interest are set to record and are

synchronized with DIC images.

 Verify triggering of test frame and DIC images.

 Verify that lights are on, exposure is correct, frame rate is correct.

4. Processing of DIC Images (5)

 Select initial correlation and user-deﬁned parameters.

 Perform initial correlation of images.

 Re-analyze images using diﬀerent user-deﬁned parameters. (For example,

do a VSG study if strain is the QOI.)

 Based on results of the diﬀerent correlations, select a ﬁnal set of user-deﬁned

parameters.

 Correlate all images using ﬁnalized parameters.

 Quantify variance and bias errors using ﬁnalized parameters (5.4).

5. Reporting Requirements (6)

 Justify and document selection of all choices in the test and analysis of DIC

data.

Design of DIC Measurement (2)

Determine the

Measurement

Requirements (2.1)

Start

Equipment Selection to

Meet Measurement

Requirements (2.2)

Pattern Selection

(2.3)

Requirements:

• QOI (2.1.1)

• ROI (2.1.2)

• FOV (2.1.3)

• Position Envelope (2.1.4)

• 2D-DIC vs Stereo-DIC (2.1.5)

• Stereo-Angle (2.1.6)

• DOF (2.1.7)

• Spatial Gradients (2.1.8)

• Acceptable Noise-Floor (2.1.9)

• Frame Rate (2.1.10)

• Maximum Exposure Time (2.1.11)

• Synchronization & Triggering (2.1.12)

• Camera & Lens (2.2.1)

• Mounting Equipment (2.2.2)

• Aperture (2.2.3)

• Lighting & Exposure (2.2.4)

• Approximate Image Scale

Adjustments

Required

Pre-Calibration Routine (3.1)

Calibration (3.2)

• Finalized Test Procedure (3.1.1)

• Cleaned Camera Detector & Lens (3.1.2)

• Warmed Up Cameras (3.1.3)

• Synchronized Cameras (3.1.4)

• Applied DIC Pattern (3.1.5)

• Veriﬁed FOV, DOF, & Locked Lenses

(3.1.6.1-3.1.6.3)

Acquire Static

Images

(optional)

Review Live or Static Images (3.1.6.4) for:

• No Glare

• Required DIC Pattern

• Proper Focus

• Sufﬁcient Contrast

• Even Lighting

• Proper Exposure

• Clean Lens & Camera Detector

• No Vibration Issues

Images

Sufﬁcient

Setup Hardware

Adjust DIC system until high-quality images are obtained

Finalized

Hardware

Setup

Mock Setup

(optional)

Yes

Adjust Camera or Lens

Figure A.1: Flow chart illustrating the main steps involved when conducting

DIC measurements in conjunction with mechanical testing of a planar test

piece (part 1).

Reporting (6)

Post-Calibration Routine (3.3)

System

Ready

Verify Calibration (3.3.2)

• Intrinsic Parameters (3.3.2.1)

• Extrinsic Parameters (3.3.2.2)

• Absolute Distances (3.3.2.3)

Perform a Final Review of System (3.3.3)

• Abbreviated Noise-Floor Analysis (3.3.3.1)

• Check for Heat Waves (3.3.3.2)

• Stability (3.3.3.3)

• etc. (3.3.3.4)

Execution of Test with DIC

Measurements (4)

Select User-Deﬁned

Parameters (5.2)

Considering Variance &

Bias Trade-Offs (5.4)

Correlate All

Images Using

Finalized

Parameters

Quantify Noise and Bias

Errors using Finalized

Parameters (5.4)

Acquire Static Images

(3.3.1.3)

Finalized User-Deﬁned

Parameters (5.2)

Yes

Adjust Camera or Lens

Acquire Rigid-Body-Motion

Images (3.3.1.5)

Review Image Quality

(see 3.3.1.4 & 3.1.6.4)

Images

Sufﬁcient

Adjust Camera or Lens

Adjust Lighting

or Exposure

Yes

Variance QOI

Bias QOI

Legend

Process

Decision

Input/

Output

Start/Stop

(Section Number)

Adjust Lighting

or Exposure

Final

Calibration

Values

Figure A.2: Flow chart illustrating the main steps involved when conducting

DIC measurements in conjunction with mechanical testing of a planar test

piece (part 2).

B — Focal Length, Field-of-View,

and Stand-Off Distance

Thin lens theory

25,26,27

deﬁnes a set of basic equations that may be used to approxi-

mate FOV or SOD between the camera(s) and the patterned test piece, for given set

of camera and lens hardware. These equations are not exact, due to intricacies of real

optical systems that are beyond the scope of the current edition of this guide, but they

have proven to be close enough to use for setting up DIC in practical situations. They

begin to break down in macro-lens photography situations, and should therefore not

be relied upon for high accuracy when the FOV size shrinks to twice the sensor size or

smaller. In general usage, they are good for determining, for example, which lens from

a kit of lenses to use, or for determining the approximate SOD.

In many typical DIC setups, the camera or sensor bar is set in a ﬁxed location

relative to a test ﬁxture containing a patterned test piece (i.e. a ﬁxed SOD), and it

is desirable to calculate the FOV for a given lens, to determine if the ROI on the test

piece will be imaged. The characteristic length of the FOV (L

F OV

) can be closely

approximated using the focal length of the lens(es) used (L

F L

), the distance from the

camera(s) to the patterned object (i.e. the SOD, L

SOD

) and the width of the camera

sensor(s) (L

F OV

= L



SOD

− L

F L



(B.1)

Online calculators

are available to quickly solve for the FOV, and extend the

formula to account for rectangular sensors. Conversely, in some setups the required

FOV is ﬁxed by the test piece, and the lens focal length is ﬁxed by the hardware on

hand, but the distance to the test piece (i.e. SOD) may be adjusted by moving the

camera or sensor bar. In this case, rearranging equation B.1 gives:

SOD

= L

F L



F OV

+ 1



(B.2)

Thin Lenses by Prof. Richard Fizpatrick: http://farside.ph.utexas.edu/teaching/302l/

lectures/node140.html

Lens (optics): Imaging properties on Wikipedia: https://en.wikipedia.org/wiki/Lens_

(optics)#Imaging_properties

Lens Focal Length Calculator by Iacopo Giangrandi: http://www.giangrandi.ch/optics/

focalcalc/focalcalc.shtml

Calculator for ﬁeld-of-view (FOV) of a Camera and Lens by Wayne Fulton:

http://www.scantips.com/lights/ﬁeldofview.html

Table B.1: Photron MH4 stereo pair at 1.52 m (60 in) SOD (L

SOD

= 1.52 m)

Focal Length (L

F L

) FOV Width (L

F OV

)

6 mm 1.22 m (48 in)

8.5 mm 0.86 m (34 in)

12 mm 0.61 m (24 in)

The camera sensor width is usually found in manufacturer speciﬁcations for the

camera hardware. Sometimes, pixel size (typically in microns) is used instead. In the

latter case, L

is simply the pixel size multiplied by the number of pixels across the

image. This is also true in cases where a cropped image, less than the full image size,

is being used: simply multiply the width of the cropped image in pixels by the physical

pixel size to obtain L

. If only the sensor width is in the speciﬁcations, it may be

necessary to ﬁrst calculate the pixel size by dividing L

by the full frame pixel width.

Note that “binning” allowed by some cameras does not reduce L

, because it still

utilizes the full sensor, but combines data from multiple pixels to create a lower pixel

count image (unless binning is used in addition to cropping of the image).

A practical use of Eqn. B.1 and Eqn. B.2 is building quick-reference tables for

ﬁelds-of-view or SODs for a given set of hardware. For example, consider a Photron

MH4 hardware kit containing two cameras, a stereo mounting bar, and several ﬁxed

focal length (non-zoom) lenses. In this case, the stereo mounting hardware ﬁxes the

stereo-angle, which eﬀectively locks the SOD at L

SOD

= 1.52 m (60 in). In the kit

are 6, 8.5, and 12 mm lenses. A chart can be drawn up for the available measurement

widths achieved by switching lenses, using Eqn. B.1 and a camera sensor width of

= 4.80 mm, reduced from the full sensor width of 5.12 mm for these cameras

to account for stereo overlap. The resulting table is shown in Table B.1. Similar

measurement width tables may be formed for other camera kits, to be used as quick-

reference guides. When the FOV is ﬁxed (constrained by the test piece size in a

universal test machine, for instance), Eqn. B.2 may be used to create quick-reference

tables of working distance required to use various lenses in a kit or being considered

for purchase.

Another practical outcome of Eqn. B.1 and Eqn. B.2 is that the FOV and SOD for

a speciﬁc lens and camera can be memorized, and then the linearity of lenses can be

invoked to quickly calculate the FOV and SOD for other lenses. For example, one could

memorize that the widely used 2/3” format camera sensor, with an 8 mm focal length

lens, has a FOV that is slightly less wide than the SOD. Therefore, with a 50 mm lens

(which has a focal length about 6X longer than an 8 mm lens), the required working

distance is approximately 6X the desired FOV.

Notes