Trembling Particle Swarm Optimization for Modified Possibilistic C Means in Image Segmentation

In this paper, a new possibilistic C-means clustering algorithm is proposed for image segment. Fuzzy C-Means isn´t better for the image with noise, and Possibilistic C means(PCM) clustering algorithm is very sensitive to initialization and parameter. In this study, in order to avoid the weakness, a modified PCM was presented. It utilizes the strong ability of the global optimizing of the tPSO Algorithm which avoids inefficiency in fine-tuning solutions and stagnation result in local optimum. Furthermore, the tPSO defines the centers and numbers of clustering automatically. Two algorithm combined to find a global optimizing clustering. the experimental result reveals the advantage of the new algorithm lies in the fact that it can not only avoid the coincident cluster problem but also has less initialization sensitivity and higher segmentation accuracy.