Open emmanuelle opened 7 years ago
I postpone this issue to 0.17
I have an implementation that @jni and I briefly discussed over on scipy. My code is simpler than the stalled PR #2739 from a few years ago, but in my implementation, the Gaussian is not rotated like it is in that PR and the current 2D implementation.
Which implementation would be better? I'd be happy to try to take a stab at finishing up the old stalled PR if consistency is preferred.
At the moment, Gabor filters are not 3-D compatible.
See http://demonstrations.wolfram.com/Gabor3D/ for the shape of a Gabor kernel in 3D, and http://www.sciencedirect.com/science/article/pii/S0262885605000934 for an example of using 3D Gabor filters