Issue using incremental correlation

[ryarosario]

Hi! I'm trying to run 2D DIC on a soft tissue sample that will be pulled to large deformations. To enable correlation at higher strains, I'd like to use incremental correlation. However, when I try to use incremental correlation, there is a lack of correlation for anything beyond the first deformed frame.

First, here's a look at the gamma output using a fixed reference

Figure 1. Frame 0 using fixed reference, REF

Figure 2. Frame 1 using fixed reference, REF

Now, here's a look at those same results using incremental correlation, showing just Frame 1:

Figure 3. Frame 1 using incremental correlation with fixed reference REF

Note there are no values for gamma anywhere in the ROI.

Lastly, to ensure correlation was possible I ran the analysis with a fixed reference, using Frame 0 as the reference and Frame 1, essentially trying to mimic what should happen with incremental correlation:

Figure 4. Frame 1 using a Frame 0 as a fixed reference

Given that there are overlapping regions that successfully correlated in Figure 1 and in Figure 4, I would expect that using an incremental correlation should produce some regions where subsets successfully correlated, but that's not what I'm seeing. I'd greatly appreciate any guidance on how to proceed.

-Ryan

[dicengine]

Hi Ryan,

Sorry you're running into trouble. How big are the deformations between REF and frame 1? It could be that the deformation exceeds the max deformation tolerance and the subsets are being disabled as a result. What is the STATUS_FLAG field for the failed subsets?

[ryarosario]

Thanks for the response! I'm seeing normal strain increments of roughly 1-2% in the pull direction between frames, with smaller shear strains and strain orthogonal to the pull direction. I have included some images below for your reference.

Figure 1. XX strain for frame 0 with REF as fixed reference

Figure 2. XX strain for frame 1 with REF as fixed reference

Regarding your second question, I'm seeing STATUS_FLAG = 24 for the failed subsets.

Please let me know if any more info would be helpful.

[dicengine]

Can you provide the params.xml file for this analysis? It would be helpful to see which initialization method you are using. Thanks

[ryarosario]

Sure thing. Here they are! I changed each to a txt file because I received an error indicating that xml files were not permitted when I tried to attach them.

params.txt
input.txt

[dicengine]

It's hard to say what's causing the problem. The status flag being 24 isn;'t very helpful, it just reiterates that the correlation failed for some reason. I was thinking it could be because the deformations exceeded the jump tolerance, but that would lead to a different status flag. Are you willing to upload the images you are having this issue with and I can try them in debug mode to see what's going on?

[ryarosario]

Sure thing, and thanks for all your help here. Here are the three images that I used for the analysis:

[dicengine]

sorry, one more thing. I need the subset_defs.txt file

[ryarosario]

Here you go! Any sorry for the late reply here. I was out on vacation beginning Thursday afternoon.

subset_defs.txt

[dicengine]

I've looked at your example for a while and can't figure out why it doesn't work with the incremental formulation. It either has to do with a bug in the feature matching initializer for incremental or ref/def image gradients not getting updated correctly after each frame. I'll need to dig deeper on this, but I'm not sure how long it will take to make progress. In the meantime, it seems like the standard formulation (non-incremental) works for this image set. Perhaps you can use that until we figure out what's wrong with the incremental.