A wavelet-based multiresolution color image segmentation

In this paper, we proposed a new image segmentation scheme based on the multiresolution image by a wavelet transform and the image segmentation by a watershed transform. At first, the original image is transform into a multiresolution image by using Haar wavelets. Then, the watershed of the lowest-resolution image is computed. To avoid over-segmentation, a fuzzy similarity criteria is introduced to merge neighboring regions. Experimental results demonstrate that the presented scheme can be applied to the segmentation of noise or degraded images as well as to over-segmentation