dicengine / dice

Digital Image Correlation Engine (DICe): a stereo DIC application that runs on Mac, Windows, and Linux
Other
354 stars 127 forks source link

Sub-pixel level Algorithms #211

Open DICeuser22 opened 3 years ago

DICeuser22 commented 3 years ago

The pixel-level algorithms used at DICe are "Zero-normalized sum squared differences (ZNSSD)" correlation criteria, but What are the sub-pixel level algorithms used?

dicengine commented 3 years ago

I'm not sure I understand your question. The ZNSSD criteria is simply a way (or norm) to measure the difference between the intensity values of two collections of pixels. I'm wondering what connection you are trying to make with sub-pixel accuracy, which has to do with many things in addition to the correlation criteria. One can certainly achieve sub-pixel accuracy with an algorithm that includes a ZNSSD correlation criteria.

DICeuser22 commented 3 years ago

Yes, the ZNSSD criteria is simply a way (or norm) to measure the difference between the intensity values of two collections of pixels, but, if compression, traction, shear etc occurs in the pixels during the deformation, what algorithms are used to take that into account?

dicengine commented 3 years ago

Those are represented in the so called "shape functions" which are used to approximate the deformation. You might find the answers to your questions in this report: https://github.com/dicengine/dice/blob/master/doc/reports/LocalGradientAlgorithm.pdf. In particular, the first page might address your curiosity about representing deformations in DIC.