Color Image Segmentation Using Optimized FCM Based on Modified Xie-Beni’s Validity Index

Abstract

Abstract

Color image segmentation is an important task for image processing. There are many techniques used for segmentation; among them clustering is the most popular technique. The main issue of clustering process is to determine the number of clusters present in the image, which often is decided by the user or by cluster validity index. It is illogical to have correct or exact knowledge about the data points present in the image to segment it. Thus, identifying optimal number of clusters presents in the image is the prime objective of this paper. In this paper, the number of colors or clusters present in color image is identified by giving proper fuzzification density to the image data points with the fuzzy membership function of FCM (Fuzzy C-means) based on Xie-Beni’s (XB) validity index by modifying it. XB validity index is very popular validity index in FCM. The proposed technique decides the optimum number of colors or cluster centers present in the color images in such a way that same group of image data points having minimum distance and separation between the cluster centers is maximized. This method solves the cluster validation problem. This prospective method is assessed on surely understood common natural color images and compared with the benchmark XB validity index.

Keywords: Color image segmentation, XB validity index, Fuzzy C-means (FCM), optimal clustering

Cite this Article

Khushbu Raval, Ravi Shukla, Ankit Shah. Color Image Segmentation Using Optimized FCM Based on Modified Xie-Beni’s Validity Index. Journal of Image Processing & Pattern Recognition Progress. 2017; 4(2):4–12p.