11. Digital Image Correlation Flashcards
What are some general points on the usage of Digital Image Correlation (DIC)
- Allows the measurement of full-field displacements - Easy to use, accurate and fast - Track patterns to determine displacements - Cost effective - Wide range of applications
What are the principals of the DIC method
- Image correlation relies upon “fingerprint” of digital grayscale patterns
- Pattern is discretized into small windows/subsets
- Displacements calculated and differentiated to obtain strains
What does DIC stand for?
Digital image correlation
DIC relies on digital patterns, what must the characteristics of those patterns be?
- High contrast - Random pattern - The scale must be relevant to deformation and the recording device - Sharp grayscale - resolution should be 1 pixel at recording device i.e. tone can change every pixel
If a pattern naturally exists on a specamine, does another pattern need to be projected for DIC?
NO!
Name 4 typical methods of creating a pattern on a sample for DIC?
- Spray painting - usually with black and white paint, with large deformation it is better to speckle the surface to avoid paint crazing.
- Abrasion of surface - for small displacements
- Chemical Etching
- Use of natural pattern on surface
Lighting plays a a role in the type of speckle pattern used
What are the assumptions involved in Digital image correlation?
- Changes are comparably small
- Changes are continuous, like in a strain field
- Two (or more) states of an object exist
- Comparison by correlation of images and/or by feature extraction
Draw a sketch of a 2D experimental arrangement for DIC
CCD camera perpendicular to the surface of the object
Random pattern on the surface of the object
White lights provide illumination
Moving the camera further from the specamine minimizes erros due to out of plane displacements
What are the basics of 2d image correlation in DIC?
- Small square subsets of an undeformed image are matched to their locations in deformed images
- typical subset sizes are 11x11 to 41x41 pixels, it can go up to 61x61 pixels
- once the first subset is analysed an overlapping subset is then analysed, this continues until the entire displacement field is analysed
- Current programms can run over 1000 subsets per second on a 500 MHz PC
Because it is possible to match undeformed square subsets with deformed non-square subsets, deformations and in-plane rotations are not a problem for this method
What are some requirements for the accurate usage of DIC?
- Initial estimate for rigid body motion
- Planar surfaces, with minimal out-of-plane motion in 2d systems
- Appropriate subset size for applied pattern
- Analysis is algorithm dependent, feature dependent - rule of thumb: 1/100 pixel
How are the initial estimates for rigid body motion in DIC made?
- The estimate is provided by a simple point and click procedure incorporated in the program
- The estimation of initial rigid body motion and strain allows direct comparison of the initial image to any other image from the test
- The program provides two different types of initial guesses
- If the location of the first point experienced mostly translations, the user need only indicate the same location in the second image.
- Multiple points are needed for image pairs with large strains.
If the images were taken at incremental load steps no initial guess is required
How do we minimize the interference of out of plane motion in 2d systems for DIC?
- By using telephoto lenses and keeping the camera as far from the specimen as possible
- Tolerance is W/Z < 0.0001 for region of interest
- W is the out-of-plane displacement of the specimen during loading
- Z is the distance from the camera to the specimen
What is the Appropriate subset size for the applied pattern in DIC methods?
- Pattern application methods have been developed for areas of interest from 0.5 mm x 0.5 mm to 200 mm x 200 mm
- There needs to be enough information in each subset to obtain a good match between the images. If the features of the pattern are large, larger subsets are required
What are the limits of an algorithm dependant DIC analysis?
each system uses slightly different methods to track the displacements of the patterns
What are the limits of a feature dependant DIC analysis?
if the features being tracked cannot be picked up by the camera or they are not random then the method won’t work
However generally speaking can accurately discern movements down to 1/10 pixel
