Effective image clustering using self-organizing migrating algorithm

Abstract

Image segmentation is an important task in computer vision. Clustering is a common image segmentation approach which divides an image into homogeneous regions, but conventional clustering algorithms such as k-means have a tendency of getting stuck in local optima. In this paper, we propose a novel clustering algorithm based on the Self-Organizing Migrating Algorithm (SOMA). In particular, we adopt SOMA Team To Team Adaptive (SOMA T3A), a recent variant of SOMA, to image clustering. Experimental results on a set of benchmark images show excellent image clustering performance, also in comparison to other state-of-the-art metaheuristics.