THE ART AND APPLICATION OF DIC by Phillip Reu

Introduction to Digital Image

Correlation: Best Practices and
Applications
INTRODUCTION
I am pleased to be able to write this new article series in Experimental Techniques.
This journal is aimed at the practicing technologist, engineer, or researcher. It
is my goal to make this series practical. I will do this by disseminating the best
current Digital Image Correlation (DIC) research available and boiling it down to
‘‘what does it mean for me in the lab?’’ Along with consultation with the foremost
practitioners of DIC and the developers who are writing the codes, I will also
draw on my experience applying DIC to challenging measurement problems at
Sandia National Laboratories. Applications at Sandia have covered a wide range Typical outdoor DIC application at Sandia,
of experiments including electron microscopes and stereo-microscopes and large- showing interlaced DIC stereo-rigs.
scale application with fields of view of tens of meters; at low speed and at high-
and ultra-high speed (1 million frames per second). As with any experimental † Early DIC References
technique, there is an ‘‘art’’ to conducting the experiment correctly, for example 1. Lucas, B.D., and Kanade, T., in Proceedings
speckling and lighting the object. As the article series progresses, I hope that of the 1981 DARPA Imaging Understanding
it will become a guide to ‘‘Best Practices’’ in the application of DIC. This guide Workshop, pp. 121–130 (1981).
will hopefully aid both the long time users and those who are new to the field: 2. Peters, W., and Ranson, W., Optical Engi-
graduate students, technologists, and others beginning to work in the field of neering 21(3):427–431 (1982).
3. Sutton, M.A., et al. Image and Vision
experimental mechanics.
Computing 1(3):133–139 (1983).
4. Bruck, H., et al. Experimental Mechanics
29(3):261–267 (1989).
DIGITAL IMAGE CORRELATION: A BRIEF HISTORY AND ‡ Sutton,M.A., Orteu, J.J., and Schreier, H.W.,
DEFINITION Image Correlation for Shape, Motion and
Deformation Measurements, Springer, New
DIC is the most important advance in experimental mechanics since the strain
York (2009).
gage. Why? Because imaging has always been important to science. Inventions
such as the telescope and the microscope that allowed scientists (a)
to ‘‘see’’ new things for the first time, invariably led to an
explosion of scientific and engineering knowledge. This is
true of DIC as well, where the deformation (strain) of an
object can be visualized. Historically, the first practical form
of DIC came about in the 1980s† because of the advent of
the digital camera with a sensor plane that is uniform and
the affordable processing power of the emergent personal
computer. The DIC concepts, however, did not appear out of
nowhere. They were a logical development of the earlier and
related photogrammetry techniques used for aerial imagery,
robotic vision, and microscopy. With the newly available digital
cameras, it was a natural next step to apply the cameras in
an experiment by first using coherent light, for holography and (b)
digital speckle pattern interferometry, and then later using an
applied painted speckle pattern.‡ Once this step was taken,

The Art and Application of Digital Image Correlation is written by Phillip L. Reu
([email protected]). He received his PhD from the University of Wisconsin-
Madison and is currently a Principal Member of Technical Staff at Sandia National
Laboratory. Phillip specializes in the development of novel full-field measurement
techniques for collecting data in previously un-measurable regimes. He began working
with digital image correlation in 2004 and has applied it in the field to a wide range of
experiments including: quasi-static, micro-scale, multi-system, large fields-of-view, and
ultra-high speed. His image correlation research is focused on understanding the effect of
the unavoidable compromises made in field measurements to the final DIC uncertainty.
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed
Martin Company, for the United States Department of Energy’s National Nuclear The cost for digital cameras has steadily decreased through time;
Security Administration under contract No. DE-AC04-94AL85000. both for machine vision (a) and high-speed cameras (b).
doi: 10.1111/j.1747-1567.2011.00798.x
© 2012, Copyright the Author
Journal compilation © 2012, Society for Experimental Mechanics January/February 2012 EXPERIMENTAL TECHNIQUES 3
THE ART AND APPLICATION OF DIC

most of the familiar techniques used in DIC have followed naturally;

subset matching, gray-level interpolation, and stereo camera calibration for
triangulation. These have developed into the three main branches of DIC: for Commercial DIC codes
in plane, 2D-DIC; for x, y, and z data, 3D-DIC, and for measurements within • ARAMIS—Trilion (GOM Gbmh)
a volume, V-DIC. The rapid adoption and spread of DIC has been greatly • ISTRA 4D (Q-4xx)—Dantec Dynamics
assisted by the phenomenal advances in imaging technologies over the last • StrainMaster—LaVision Inc.
20 years, including high-resolution cameras (14 megapixels are now common) • Vic3D—Correlated Solutions
and high-speed cameras (5000 to millions of frames per second). Furthermore,
the great flexibility of the mathematical concepts underlying both 2D- and 3D- University codes (short list)
DIC enable it to be used with methods that are not imaging in the traditional • MatchID—Catholic University College
optical sense (e.g., atomic force microscope [AFM] applications, where the image Ghent, KULeuven
is normally a height map; scanning electron microscope [SEM] applications; and • Opticist—The Catholic University of Amer-
X-ray and computed tomography [CT]). ica
• Matlab—Karlsruhe Institute of Technology
(KIT) and Johns Hopkins University
SOME CHALLENGES WITH WRITING THE ARTICLE
The goal is to make this series as general as possible; however, references to
various commercial and university software codes will be necessary at times.
I see the greatest hurdle being that many of the commercial codes are now
functioning as ‘‘black boxes’’ where the images are put in and results are cranked
out. This is valuable for the user, but without input from the software companies †
Hild, F., and Roux, S., Strain 42(2):69–80
important aspects of the process (i.e., recommended calibration methods, subset (2006).
sizes, and speckle patterns) can be difficult to discuss. I believe that many of ‡ [email protected] is my contact
the best practices will be universal to all the codes because all the underlying e-mail.
principles are similar. Therefore, a good speckle pattern for one code should be
the same for all codes, whether subset based or not. At a fundamental level,
this is true because the quality of the correlation is a function of the quantity of The ‘‘hidden components’’ of DIC
1. Calibration (3D and maybe 2D)
information contained in an image. An image with a large quantity of information
2. Subset shape function (both 3D and 2D)
will render better results than one with low information content. Information in
3. Gray level interpolation (both 3D and 2D)
this context is defined by the contrast and number of edges in an image.
4. Subset matching (both 3D and 2D)
5. Triangulation (3D only)
Recently, a new type of DIC software based on a finite element approach is 6. Post-processing (both 3D and 2D)
being developed that solves the entire field of view simultaneously,† rather than
the subset approach used by the earliest researchers and the commercial codes.
Notwithstanding, many of the principals involved here will also apply to full-field
DIC methods (e.g., optimum speckle).

CONCLUSION
With this introduction, I would like to invite you to join with me in developing
‘‘DIC Best Practices.’’ Please e-mail me article ideas and questions you would like
answers to.‡ I will consult with other experts in the field and attempt to present
practical answers to your questions to improve everyone’s DIC capabilities.
Together we can make this a community forum where great ideas can be shared
to improve and extend the use of DIC.

I would also like to direct the reader’s attention to an excellent book on DIC
published by Sutton, Orteu, and Schreier, titled Image Correlation for Shape,
Motion, and Deformation Measurements. It is available from Springer press.
Many concepts presented here will be drawn from this book.‡

The next articles in this series will introduce the ‘‘Hidden Components’’ of DIC.

4 EXPERIMENTAL TECHNIQUES January/February 2012