Watershed segmentation with boundary curvature ratio based merging criterion

This paper proposes to incorporate boundary curvature ratio, region homogeneity and boundary smoothness into a single new merging criterion to improve the oversegmentation of marker-controlled watershed segmentation algorithm. The result is a more refined segmentation result with smooth boundaries and regular shapes. To pursue a final segmentation result with higher inter-variance and lower intra-variance, an optimal number of segments could be self-determined by a proposed formula. Experimental results are presented to demonstrate the merits of this method. KEY WORDS Watershed segmentation; merging criterion; boundary smoothness; optimal number of segments

• Type: Conference
• Year: 2007
• Where: SIP
• Authors: X. He, N. H. C. Yung, K. P. Chow, Francis Y. L. Chin, Ronald H. Y. Chung, Kwan-Yee Kenneth Wong, Kenneth S. H. Tsang
• Keywords: Boundary Smoothness, Image Processing, SIP 2007, Watershed Segmentation, Watershed Segmentation Algorithm