Closed mayukh18 closed 8 years ago
I think that segmentation might be more appropriate for a higher level toolkit like scipy or opencv. We don't really have anything that's related to computer vision, other than the low level tools to deal with image formats.
I'd like to see the ability to segment by color. Right now it's pretty hard to build masks...
Will it be a good idea to contribute a segmentation method in ImageOps? My code works best on grayscale or each layer of color and is implementation of fuzzy c-means clustering. It can work on a total RGB or CMYK image too, but with some color inaccuracies if we choose very few color clusters or segments.