Hello, currently I am working on image proccesing of brain MRI scans. I am trying to use Otsu's multi thresholding method to separate objects before using Watershed segmentation. Trouble is that the algorithm often uder or over segments the picture. Is there a way how to determine optimal number of thresholds? Its varying from image to image in range of 4-7 with median 5. Simply running the whole algorithm to test for the best result would be too time costly.