0
Review Article

Computer Vision-Based, Noncontacting Deformation Measurements in Mechanics: A Generational Transformation

[+] Author and Article Information
Michael A. Sutton

Carolina Distinguished Professor
Department of Mechanical Engineering,
University of South Carolina,
300 South Main Street,
Columbia, SC 29208
e-mail: Sutton@sc.edu

Image inaccuracies were only apparent when comparing regions of the images using image correlation–based image analysis methods that have subpixel measurement accuracy. This was also found to be true with most advanced measurement systems, including imaging systems such as SEM, AFM, optical microscopy, and confocal microscopy.

In fluid mechanics, the following convention is becoming increasingly popular to describe measurement methods. Methods are qualified as nDmC, where n is the Dimensionality of the support (region being interrogated) and m the number of Components being measured. In this convention, 2D-DIC would be 2D2C, 3D-DIC would be 2D3C, and V-DIC would be 3D3C.

The concepts underlying the method are based primarily on continuum mechanics principles presented by Professor Donald E. Carlson (deceased) in his theoretical classes on continuum mechanics in the Theoretical and Applied Mechanics Department (merged with Department of Mechanical Engineering) at the University of Illinois, Champaign-Urbana.

The intensity field shown in Figs. 3 and 4 have random variation across each subset, providing the basis for automating the search process across the entire image.

Technology developed by the author and his colleagues was spun off in 1996 to form the only US-based corporation focused on development and sales of DIC-based measurement systems, Correlated Solutions, Incorporated, 121 Dutchman Blvd, Columbia, SC 29210, www.correlatedsolutions.com.

Technology developed by the author and his colleagues was spun off in 1996 to form the only US-based corporation focused on development and sales of DIC-based measurement systems, Correlated Solutions, Incorporated, 121 Dutchman Blvd, Columbia, SC 29210, www.correlatedsolutions.com.

Manuscript received January 29, 2013; final manuscript received July 3, 2013; published online August 29, 2013. Assoc. Editor: Ellen Kuhl.

Appl. Mech. Rev 65(5), 050802 (Aug 29, 2013) (23 pages) Paper No: AMR-13-1009; doi: 10.1115/1.4024984 History: Received January 29, 2013; Revised July 03, 2013

The remarkable increases in computational speed and memory size that have occurred over the past three decades have impacted virtually every area of scientific inquiry. With regard to the measurement community, the digital age has revolutionized the way in which data are acquired, stored, and analyzed to extract the maximum amount of information. For the purposes of this discussion, the focus will be on computer vision–based, noncontacting measurement methods, specifically those commonly known as 2D image correlation, 3D image correlation, and digital volume correlation (also known as volumetric image correlation), and their increasingly significant role in the broad field of solid mechanics. The review closes with a visionary perspective regarding the integration of measurements and models and the impact it may have on the design process.

Copyright © 2013 by ASME
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Fig. 1

Flow chart for the methods discussed

Grahic Jump Location
Fig. 2

Speckle shearing interferometry fringe pattern [1] with fringes defining lines of constant out-of-plane slope, ∂w/∂y. (Image circa 1979 obtained in TAM Department at the University of Illinois.)

Grahic Jump Location
Fig. 3

Schematic of the mapping process for the intensity pattern, I(x), as the specimen is deformed

Grahic Jump Location
Fig. 4

Schematic of a 2D-DIC system including (a) digital camera with lens to image specimen, (b) storage system (personal computer (PC) with digitizer), (c) lighting system to illuminate specimen. Surface coated with high-contrast speckle pattern.

Grahic Jump Location
Fig. 5

Image-matching process to obtain specimen displacements. Image subsets in undeformed image are individually compared to deformed subsets to identify optimal results. Subset size determined by speckle pattern to ensure good contrast so that matching process is accurate.

Grahic Jump Location
Fig. 6

Standard deviation in displacement versus normalized measure of intensity pattern noise

Grahic Jump Location
Fig. 7

Standard deviation in displacement versus normalized square of gradients in intensity pattern

Grahic Jump Location
Fig. 8

Displacement bias versus subpixel part of motion for various levels of intensity pattern noise

Grahic Jump Location
Fig. 9

Measured vertical strain fields using an AFM (a) uncorrected strain measurements; (b) corrected for time-varying drift; (c) corrected for both time-varying drift and spatial distortion. Results in (c) indicate that most of the image distortions, but not all, have been removed. (Graphs courtesy of Professor Xiaodong Li, University of South Carolina.)

Grahic Jump Location
Fig. 10

Correlation coefficient field of DIC analysis of the AFM topographical images of gold coating surface subjected to different normal wear loads: (a) 0 nN, (b) 133.16 nN, (c) 209.26 nN, (d) 285.35 nN, (e) 361.44 nN, (f) 437.53 nN. Note the size of all images is 5 × 5 μm with a 2 × 2 μm worn area in the center; a high coefficient reveals where wear occurs [64]. (Graphs courtesy of Professor Xiaodong Li.)

Grahic Jump Location
Fig. 11

Photograph of original 2D-DIC system including (a) PC with monitor for image storage and viewing of crack tip region, respectively, (b) PC for image storage and image viewing, (c) three-axis translation system for camera positioning, and (d) fiber optic illuminator for on-axis illumination of specimen mounted in tensile load frame.

Grahic Jump Location
Fig. 12

Measured crack tip opening displacement (CTOD) results at two locations behind tip for several load levels. 2D-DIC results have incredibly small variability, on the order of 40 nanometers (200 atoms), delineating clear trends in the data.

Grahic Jump Location
Fig. 13

Speckle images (in the background) overlayed with Q4 meshes in the reference (a) and deformed (b) configurations of a tensile loading experiment on A533 steel. The deformed shape on the right is obtained by a global Q4-DIC analysis. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 14

Top figure (a): Vertical component of the displacement field (expressed in pixels) obtained from Q4-DIC (ℓ=16 pixels). The physical size of one pixel is 68 μm. Bottom figure (b): Correlation residuals expressed in gray levels (image quantization is 8 bits) for the measured displacement field. The effect of the crack in the top left corner is visible in the inset image. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 15

Top figure (a): Map of the damage variable D for the last step of loading. One clearly sees in the left-hand top corner conditions that led to the initiation of a major crack, eventually leading to failure of the sample. Bottom figure (b): Identified damage law. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 16

Displacement fields expressed in nanometers. Top figure: x-direction; Middle figure: y-direction; Bottom figure: z-direction. Measurements obtained by using global DIC with AFM images of a Double Cleavage Drilled Compression experiment on glass [110]. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 17

Modern stereovision system. Light source is located between the cameras. Cameras are mounted to a rigid crossbar and can be rotated and translated along bar. Bar is mounted to a sturdy tripod so that the height can be adjusted as needed.

Grahic Jump Location
Fig. 18

Schematic of pinhole projection imaging model used in stereovision. On the right are several views of a typical calibration grid with an array of equally spaced markers. Typically 50 or more rotated and translated positions of the grid are used for calibration of the cameras in the stereovision system.

Grahic Jump Location
Fig. 19

Prediction flow chart for variance in 3D position

Grahic Jump Location
Fig. 20

Comparison of experimental results and predictions for strain εxx

Grahic Jump Location
Fig. 21

Schematic with photo of the experimental setup. Sand-filled pit used as blast medium. Two high-speed cameras mounted outside pit and aligned to view specimen. One gram of pentaerythritol tetranitrate explosive is buried below specimen and detonated. Rectangular Al-6061 specimen bolted into steel frame for experiments.

Grahic Jump Location
Fig. 22

Measured out-of-plane accelerations [127] in central region of plate during the blast loading process. L-R on top row: t = 16.25 μs, 32.5 μs, 48.75 μs. L-R on bottom row: t = 65 μs, 81.25 μs, 97.5 μs. Full field data of this type provides clear trends in the specimen's dynamic response that is used to improve understanding of high-rate events and validate simulation model predictions.

Grahic Jump Location
Fig. 23

Schematic of confined masonry wall retrofitted using horizontal aluminum strips embedded in the bed joints and anchored into the confining reinforcing columns at top and bottom. Cyclic shear displacement is applied along top surface by hydraulic ram using vertical steel backbone for support. (Photograph courtesy of Professor Fabio Matta at the University of South Carolina.)

Grahic Jump Location
Fig. 24

Close-up of random pattern applied to wall surface with sampling resolution of 0.052 in/pixel (1.33 mm/pixel). Tape measure is in inches. (Photo courtesy of Professor Matta.)

Grahic Jump Location
Fig. 25

Experimental setup with diode light arrays shown on right and left sides. (Photograph courtesy of Professor Matta.)

Grahic Jump Location
Fig. 26

(a) maximum tensile strain contours shown as dark and light, irregular shaped bands extending from bottom left to top right when displacement to left is 85% of maximum value, (b) maximum tensile strain contours shown as dark and light, irregular shaped bands extending from top left to bottom right when displacement to right is 85% of maximum value, (c) photograph of the final fracture pattern after specimen failure. Red corresponds to + 10,000 με. (Courtesy of Professor Fabio Matta.)

Grahic Jump Location
Fig. 27

Photograph of the stereoscopic system used in the cylinder shock loading experiment. Reference mirrors in which the part and the calibration target can be seen. Above the mirrors, the pyrotechnic flashes are located in wood cases. (Photo courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 28

Left: One of the reference images in which the observed object and the calibration targets can be observed. Right: Deformed shape at the end of the analysis. The deformed mesh represents the 3D reconstruction, while the points represent the interpolated surface. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 29

Left: Spherically indented cylindrical aluminum foam specimen imaged by micro-CT scanning. Right: Maximum compressive strains obtained using volumetric DIC. Results show maximum compressive strain is below the top surface. (Graphics courtesy of Professor B. K. Bay, Oregon State University.)

Grahic Jump Location
Fig. 30

Top: Color contours for longitudinal component of the displacement field (in micrometers) for the maximum load level after 45,000 cycles with an X-DVC approach. The mean rigid body motion is subtracted. Bottom: X-FEM and X-DVC mode I stress intensity factor along the crack front. Experimental results are compared with numerical simulations (X-FEM) on the same mesh, the same crack geometry, and under experimental boundary conditions. (Graphics courtesy of Professor F. Hild, ENS-Cachan in France.)

Grahic Jump Location
Fig. 31

Conceptualized flow chart for fully integrated experimental-numerical methodology that embodies the next generation design concept

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In