Image segmentation using clustering with fireworks algorithm

Abstract

This paper presents a hard clustering technique using fireworks algorithm with adaptive transfer function (FWAATF) for image segmentation. The fireworks algorithm (FWA) is a recently developed new Swarm Intelligence (SI) algorithm for function optimization. This algorithm simulates the process of fireworks explosion in the night sky. The main characteristic of FWA is the good balance between exploration and exploitation during the search process. The exploitation is done using good fireworks whereas the bad fireworks are responsible for exploration. FWA shows its efficiency and effectiveness in numerical function optimization over other SI algorithm like particle swarm optimization (PSO). FWA-ATF is a modified version of basic FWA and in this work, it is used in hard clustering technique to segment the image. FWA-ATF is used to find the optimal cluster centroids corresponding to different regions in the image. The proposed clustering technique is applied to segment four benchmark images and the well-known cluster validity index-Dunn's Index is used to measure the performance of the proposed clustering technique quantitatively. The performance of the proposed method is compared with clustering using K-means, PSO and basic FWA. The experimental results demonstrates that the proposed clustering technique with FWA-ATF performs better than other methods in segmentation for most of the images.