An improved watershed transformation for image segmentation
Ren Ming-yi, Xiaofeng Li, LI Zai-ming · 2008
In this paper, an improved watershed algorithm for image segmentation is proposed. First the morphological opening/closing reconstruction is fulfilled, then the morphological gradient image is calculated, and the gradient image is nonlinear transformed based on vision characteristics of human eye, last the watershed transformation is computed and the image is segmented. The simulation results show that this improved watershed algorithm can greatly alleviate the over-segmentation problem of the traditional watershed.