Reputation: 95
I was wondering if there was a simple python toolkit for region-based image segmentation. I have a grayscale image, and my goal is to efficiently find a complete segmentation such that the pixel values in each region are similar (presumably the definition of "similar" will be determined by some tolerance parameter). I am looking for an instance segmentation where every pixel belongs exactly one region.
I have looked at the scikit-image segmentation module (https://scikit-image.org/docs/dev/api/skimage.segmentation.html), but the tools there didn't seem to do what I was looking for. For instance, skimage.segmentation.watershed looked attractive, but gave poor results using markers=None.
Upvotes: 0
Views: 6913
Reputation: 799
The flood fill algorithm from scikit-image seems close to what you want, has a tolerance parameter as well.
For more fine-tuned control you can check out OpenCV
Upvotes: 1