Digital Image Correlation

1
Digital Image Correlation

• Janice Mochizuki
• University of Illinois, Urbana-Champaign
• CEE 498
• Professor Dan Kuchma

2
Contents

• Commercially available
• Examples of Hardware/Software
• Do It Yourself
• Finite Element Formulation
• Matlab Formulation

3
Correlated Solutions

• Company started in 1988 to commercialize DIC
  invented at the University of South Carolina
• Uses white speckle correlation
• Instruments can process real-time data at gt50 Hz
• Synchronizes with load data acquisition
• Strain accuracy from 500 microstrain to 500
  strain
• Can use on specimens from 1mm10 meters

4
Vic-2D

• Measures in-plane displacements and Lagrangian
  strain tensor
• No calibration needed
• No lasers required
• Available for analog output
• Current applications
• fracture mechanics
• bi-axial tension tests
• determination of strain location around cracks
• FE model verification

5
Vic-3D

• Measures actual object movement and Lagrangian
  strain tensor at every point
• Calibration needed for strain measurements
• Add on modules available
• Current applications
• fuselage panel testing
• car crash testing
• sheet metal forming
• deformation measurement of tires at high speeds

6
Vic-3D Vibration Analysis

• Uses Videostroboscope add-on module
• Measures all 3 displacement axes
• Only added mass is from the speckle spray

7
Vic-3D High Speed

• Uses multi-channel ultra-high speed cameras
• Acquires at 1000 to 100,000 frames/sec
• Implements rotating mirrors, photo-multipliers,
  and fiber-optics
• Requires calibration due to distortion
• uses distortion correction technology
• Example graphics impact on plexiglass panel

Shear Strain during impact
Deformation after impact
8
Costs

• Software only Vic-2D 12, 000
• Complete Vic-2D system (software, cameras,
  tripod, lights) 18,000-22,000
• Complete Vic-3D system (includes set-up and
  training) 50,000-70,000
• Vibration add-on additional 5,500
• High Speed Camera additional 30,000-200,000

• Strain Overlay (Vic-3D)

9
Dantec Dynamics

• Q-400 (3D)
• Specimens up to 400x300mm2 typical (claim to make
  special order for larger specimens)
• Measures up to 100 strain and displacements
  1/10- 1/100 of a pixel
• Optional high speed camera for 500-10,000
  frames/sec
• More user input, needs calibration
• Applications
• Material testing, fracture mechanics, dynamic
  measurements
• 75,000-95,000 (includes illumination, cameras,
  and software)

• Mapping of 3D displacement vector

10
Dantec Dynamics

• Q-450-II
• High Speed Image Correlation for Vibration
  Analysis
• More memory for cameramore time or fps recorded
• 2 seconds at 1000 fps with 4GB
• 105,000-135,000 (no illumination)
• Additional costs for illumination
• 4,000 for regular light source
• 40,000-50,000 for pulsed laser

• Membrane displacement graphic

• Out-of-plane deformation due to shock excitation

11
Finite Element Formulation

• Used to compute 2-D displacement field
• Traditional Subset-based Image Correlation
  Method (comparison)
• Measures full-field displacements with subpixel
  accuracy from pictures by matching undeformed
  with deformed pictures
• Uses least-squares algorithm to match the
  pictures

• Undeformed

• Deformed

12
Finite Element Formulation

• Element Based ICM
• Discretizes image into segments and elements
  linked by nodes
• Acquires nodal displacements by minimizing Least
  Square correlation coefficient between all the
  elements in undeformed and deformed
• Displacements at a point in an element are
  functions of displacements of element nodes
• Uses Q4 elements
• Note at the boundaries of test area, nodal
  displacements are higher since lack of adjoining
  elements to impose continuity constraints

• Deformed

• Undeformed

13
Finite Element Formulation

• Pros of FE
• FE satisfies requirements of displacement
  continuity and derivative continuity among
  elements
• FE produces smaller, more accurate strain
• Cons of FE
• Subset-based method produces smaller variations
  in displacement data

14
Matlab Formulation

• Code created by students at John Hopkins
  University in Kevin J. Hemkers group
• Created for others to build off of as an
  alternative to expensive commercial
  hardware/software
• Available online at http//www.mathworks.com/matl
  abcentral/files/12413/content/Correlation_Guide_2.
  htm

15
Matlab Formulation

• Pre-processing
• Save all photographs as .tif 8 bit greyscale
  format and number consecutively

• Chose all pictures to compare in pop-up window
• On base picture (undeformed) define a grid for
  the markers
• Chose bounds of area to analyze
• Tell program how many pixels you want between
  markers in x- and y-directions

16
Matlab Formulation

• Correlation
• Program runs through all images and markers
• Initial picture has green markers
• Plots red markers on second picture where
  displacement occurs
• Then continues with all pictures in the same
  manner tracking the markers as it goes
• Running the correlation takes about 200 seconds
  for 100 images with 200 markers

17
Matlab Formulation

• Post-Processing
• Must be knowledgeable about displacement fields
  and strains
• Delete markers with bad results
• Bad lighting
• Physical movement out of ordinary
• Not enough data found
• Option to save .jpg of each picture step to make
  a movie!

18
Matlab Formulation

• Example output interface

• Example of 3D output graph

19
Sources

• Dantec Dynamics sales representative
  http//www.dantecdynamics.com/Default.aspx?ID24
• Correlated Solutions sales representative
  http//www.correlatedsolutions.com/index.php
• Matlab http//www.mathworks.com/matlabcentral/fil
  es/12413/content/Correlation_Guide_2.htm
• Applied Optics (Finite Element)
  http//ao.osa.org/DirectPDFAccess/8A9358ED-BDB9-13
  7E-C444F5205BBAB4B1_86402.pdf?da1id86402seq0
  CFID22809832CFTOKEN13034717

20
Questions?
Have a fun Spring Break!