A Comparative Study on Image Segmentation Based on Artificial Bee Colony Optimization and FCM

The goal of image segmentation is to cluster the pixels of an image into several regions. This article describes the method of image segmentation using Artificial Bee Colony Optimization (ABC). This optimization technique is motivated by intelligent behaviour of honey bees and it provides a population based search procedure. In this article Gaussian Mixture Model (GMM) is used and each pixel class is represented by a single Gaussian function and a mix of Gaussian functions is used to segment the gray image by approximating the image histogram. The parameters of this model are estimated by ABC. Intersecting point of the gaussian functions is considered as the threshold point. The optimization technique is compared with the popular Fuzzy C Means (FCM). The proposed algorithm is found efficient over FCM. The experiment has been done over various gray scale images and segmentation of such images is very difficult due to low contrast, noise and other imaging ambiguities. The results are proved by both quantitative and qualitative measures. Keywords— Image segmentation , Gaussian Mixture Model(GMM), Fuzzy C Means(FCM), Artificial Bee Colony Optimization(ABC), Medical & Satellite image segmentation, Cluster validity index.