Reduction of over segmentation in JSEG using canny edge detector

摘要

An improved version for the JSEG color image segmentation algorithm has been presented in this paper integrating the presence of edge in canny edge detector. One major inherent problem in JSEG is that it suffers from over segmentation which ultimately reduces the segmentation quality. The focus is given to reduce the over segmentation problem in JSEG by examining the presence of canny edge between every two adjacent segment created in JSEG which matches human perception. Experiments on natural images show improved result.