Digital image segmentation method

摘要

The method involves combining together individual adjacent pixels or image segments, or modifying preformed segments by separating pixels or partial image segments, when a characteristic or several characteristics of adjacent pixels or image segments are stuffed with given homogeneity criteria as matching or not matching respectively. A tolerance is derived as the homogeneity criterium for each characteristic, within which the characteristic values of the two pixels or segments in question must be different. The segmentation is repeated iteratively over all pixels or segments.