Semi Supervised Image Segmentation Using Optimal Color Seed Selection

Abstract—Image Segmentation refers to the process of partitioning the input image into several disjoint regions with similar characteristics such as intensity, color, and texture, shape etc. Semi supervised image segmentation is clustering the pixels of an image with some prior information or constraints. The Existing semi supervised method takes EM algorithm with mouse clicks as prior information. The drawback of EM algorithm is that it is prone to local maxima problem. Because of this reason the segmentation results will not be proper for certain kind of images. In this paper a new approach of optimal Semi Supervised Image Segmentation using Genetic algorithm is discussed. The optimal seeds are obtained and passed to EM algorithm. The Optimal seeds are nothing but color centers. The nearest colors are grouped together. The color classes are given in prior and the image is clustered using EM Clustering. In this paper Genetic algorithm is applied for finding optimal color classes so that the colors in the image are clustered sharply. Natural image data set from BSD images are taken and tested. The results of the proposed method are compared with Standard EM algorithm. The results show that the segmentation accuracy is improved.