I'll come out in favor of a required wavelet dependency again here. We need that domain. Wavelet coefficients are features (should be present in skimage.feature), their coefficients sparsely describe images which is useful both for compression and denoising (skimage.restoration), and that's just off the top of my head. There are immediate uses we just can't add until we can depend on wavelet functionality. IMHO of course. Josh On Thursday, June 16, 2016 at 3:32:44 PM UTC-5, François Boulogne wrote:
Le 16/06/2016 à 18:18, Sira Ferradans a écrit :
Hello, I have one implementation question now: In order to implement the scattering, we need the wavelet transform, with morlet filters. I saw that scikit-image has support for pywt,
Not yet afaik. We have two main PR discussing this: https://github.com/scikit-image/scikit-image/pull/1837 and https://github.com/scikit-image/scikit-image/pull/1833
In #1837, @JDWarner voted in favor of pywavelet for 0.13. I would also be happy to see these features coming in 0.13; I convinced Ralf to put pywavelet on a new floating boat a couple of years ago, and scikit-image was a motivation for me.
IMO, this could be also a killing feature for users who still hesitate with scikit-image :)
-- François Boulogne. http://www.sciunto.org GPG: 32D5F22F