pderian / pytyphoon Goto Github PK

According to scipy's documentation, ndimage.imread has been deprecated since v1.0 and removed in 1.2.0.

Can you please update the code to not use it?

Hello @pderian and thanks for sharing the code of your research!!

I was wondering if you can provide a minimal example for 3d dense optical flow estimation

I am curious to apply your approach in volumetric medical images. Do you think it is possible?

Best regards,

N.A.

Salut Derian!

Thanks so much for providing this code! I've been using it to get my head around a lot of concepts in your thesis and its helped a lot.

I've used the connection coefficients function to compute the Laplacian regularization and have gotten nice results.

I had a question regarding the test of the regularization schemes that you have coded in the demo (L1 & Horn Schunk). My question is regarding the computation of the regularization term in physical space versus wavelet space and the comparison of their values with respect to the formula for the laplacian:

Are you expecting exact agreement between these two terms for every level of the transform, or is some error acceptable? Additionally, do you expect this agreement to get much better for a wavelet basis with a much higher number of vanishing moments?

Thanks in advance!