Hi all,
Do you know what are the available techniques used to determine the average speckle size found on speckle patterns prepared to conduct digital image correlation (DIC)?
For example, I have found this, where auto-correlation and radial averaging are used:
https://uk.mathworks.com/matlabcentral/fileexchange/94765-speckle-autocorrelation
Which other techniques are you aware of? I am mainly interested on the techniques/procedure/algorithms rather than the tools (software).
Best wishes,
Gustavo
Hi dear Quino
I know a few things that I will mention. I hope it would be useful.
To match the reference and deformed images, DIC tracks features on the sample surface that collectively form the
speckle pattern . Occasionally, a sample’s surface will inherently have features that suffice for a natural speckle pattern, but typically an artificial speckle pattern must be applied to the sample. The quality of DIC results are strongly dependent on the speckle pattern, and optimum speckle patterns meet the following conditions.
1.The pattern covers the sample surface in the area of interest.
2.The features that comprise the pattern (the speckles) are random in position but uniform in size.
3.The pattern moves and deforms with the sample, but does not exert a significant mechanical stress on the sample. In other words, the pattern is fully adhered to the sample, but deforms extremely easily compared to the sample. Although deformable speckles are ideal, the speckles can be rigid features that move with the sample’s deformation but do individually deform with the sample. The use of rigid speckles instead of deformable speckles adds relatively small errors to a DIC calculation (Barranger, et al.
doi:10.1111/j.1475-1305.2011.00831.x).
4. The speckle size is at least 3-by-3 pixels to avoid aliasing (Reu. doi:10.1111/ext.12111 ), but not much more than 7-by-7 pixels to achieve a relatively high density of DIC points (Reu. doi:10.1111/ext.12110 ). If speckles are much larger than 7-by-7 pixels, then there will be relatively few DIC data points possible. Also, note that these speckle sizes are not averages,but are rather the range of the smallest and largest speckles (Reu. doi:10.1111/ext.12110 ).Here is an example of a calculation to estimate the desired speckle size range: 12 mm / 2048 px * (3 to 7 px per speckle) = 18 to 41 microns per speckle.
5.The pattern has good gray scale contrast, which reduces error (Sutton, Orteu, Schreier. doi:10.1007/978-0-387-78747-3). One way to visualize this contrast is a histogram: with thenumber of pixels plotted with respect to grayscale level, the pattern has a mix of dark andbright pixels, indicated by two peaks in the histogram’s spectrum, and the separationbetween the two peaks is broad. Ideally, the two peaks look like a bimodal Gaussiandistribution.
6. The edges of the speckles are softened (rather than sharp and distinct with the background).This allows the pixels of the camera sensor to avoid aliasing the speckle edge (Reu. doi:10.1111/ext.12139).
7. The pattern is stable in the testing environment. For example, for a high-temperatureexperiment, the pattern does not decay or darken under heating.
8. The pattern has a speckle density of about 50%. When the pattern has either too few or too many speckles, then this results in features that are both too big and too small (Reu. doi:10.1111/ext.12110).
Hello Gustavo.
I will indicate some articles that supported me both in the laboratory experiences and in programming the correlation codes, since I do not use commercial software. They follow:
Here are presented several techniques for applying the texture pattern on the specimens. This pattern needs to have randomness and contrast, since the area of interest of the reference image will be sought in the area of interest of the deformed image throughout the tests. I agree here that the spray application technique with white and black paint is more common, but in metallic specimens there is an expectation of a better texture pattern with the application of black pilot and white drops. DOI 10.1007/s11340-017-0283-1
To find the correlation of image gray intensities, I recommend this article: Study on subset size selection in digital image correlation for speckle patterns. Where the mathematical forms for correlation are presented. This you can apply using Matlab for example.
Another way is using CPCorr and adjusting its subset tentatively, based on the resolution of your image, for example. Mathematical forms of correlation will always search within the image matrix for the smallest correlation.
I hope I have contributed something.
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress