FUZZY CLUSTERING FOR COLOR IMAGE SEGMENTATION USING GRAVITATIONAL SEARCH ALGORITHM

Abstract

Clustering is a key activity in numerous data mining applications such as information retrieval, text mining, image segmentation, etc. This research work proposes a clustering approach, Fuzzy-GSA, based on gravitational search algorithm (GSA). In the proposed Fuzzy-GSA approach, a fuzzy inference system is developed to effectively control the parameters of GSA. The performance of the Fuzzy-GSA algorithm is evaluated against four benchmark datasets from the UC Irvine repository. The results illustrate that the Fuzzy-GSA approach attains the highest quality clustering over the selected datasets when compared with several other clustering algorithms namely, k-means, particle swarm optimization (PSO), gravitational search algorithm (GSA) and, combined gravitational search algorithm and k-means approach (GSA-KM) In this paper, we propose a new hybrid approach for image segmentation. The proposed approach exploits fuzzy GSA for clustering image pixels into homogeneous regions. In order to improve the performance of fuzzy clustering to cope with segmentation problems, we employ gravitational search algorithm which is inspired by Newton’s rule of gravity. Gravitational search algorithm is incorporated into fuzzy GSA to take advantage of its ability to find an optimum cluster center which minimizes the fitness function of fuzzy GSA. Experimental results show effectiveness of the proposed method in segmentation different types of images as compared to classical fuzzy Algorithm