On Sun, Jun 23, 2013 at 4:16 AM, Dan <dkin@walla.co.il> wrote:
Hi,
I wish to perform first (histogram based mean, stdev, smoothness, skewness, uniformity and entropy) and second order (GLCM based contrast, correlation, energy, homogeneity) statistical texture features of an image. is it possible in scikit-image?
If so a small script will be a huge help.
Thanks.
Hi Dan, Sorry for the delayed reply. GLCM is implemented, as shown in this example: http://scikit-image.org/docs/dev/auto_examples/plot_glcm.html Unfortunately, only a handful of statistics are currently calculated. It's not too difficult to calculate your own since you have access to the GLCM. More statistics should be added to the base implementation, though. As for first order statistics, I'm not completely sure what you mean. If you just want to calculate these on a given image patch, there are some numpy functions for a few of these (`np.mean`, `np.std`), but for others it would be fairly simple to implement your own (sorry, I don't have it in me to provide an example---it's been a long week). For local statistics, maybe you're looking for rank filters: http://scikit-image.org/docs/dev/auto_examples/applications/plot_rank_filter... This only implements a few of the filters you'd want (and the entropy filter is a bit broken since it should really return float values---the output resolution/range is pretty low at the moment). Hope that helps, -Tony