dicengine
Stereo DIC: feature matching does not produce enough keypoints for color images
Hello there,
previous issue : https://github.com/dicengine/dice/issues/192
Yesterday, I had a problem performing a calibration using pictures with a resolution of 4752 x 3168. I resized them to 1920 x 1280 and now the calibration works just fine.
I trying to run an analysis of an experiment using the same equipment in the same configuration. (I also resized the pictures of the test to matched with those of the calibration). VIC3D is able to run the analysis. However, an error appears with DICe saying that not enough features are matched... I'm trying to play also with the ssig threshold but the same error appears If someone could help me with that, I would appreciate it!
Please see attached a link with the calibration and the test pictures and some screenshots.
Thanks!
https://drive.google.com/drive/folders/1dMGbOyrIwYE0dxDfDDURODjA75MRj9Al?usp=sharing
A couple observations about your images: They appear to be three channel color images. I would recommend converting them to single channel grayscale before using them in DICe. For most of DICe, the images get converted to grayscale (which can slow things down), but we don't have a lot of testing for color images. I would convert them as a pre-processing step to be safe. The second observation is that your speckle pattern is only in a portion of the image. I would make sure to define a region of interest that includes only the speckled portion of the image. See if either of those help. The conversion to grayscale may fix the feature matching issue.
I noticed on your screencapture that the correlation failed at image 14. In your folder 'resized_files_test' image nr. 14 is a calibration image (as is the case for images 15-26). I tried correlating the first 13 images without any problems.
