INVESTIGATIONS ON FUZZY THRESHOLDING BASED ON FUZZY CLUSTERING

Thresholding, the problem of pixel classification is attempted here us ing fuzzy clustering algorithms. The segmented regions are fuzzy subse ts, with soft partitions characterizing the region boundaries. The val idity of the assumptions and thresholding schemes are investigated in the presence of distinct region proportions. The hard k means and fuzz y c means algorithms have been found useful when object and background regions are well balanced. Fuzzy thresholding is also formulated as e xtraction of normal densities to provide optimal partitions. Regional imbalances in gray distributions are taken care of in region normalize d histograms. (C) 1997 Pattern Recognition Society. Published by Elsev ier Science Ltd.