Image segmentation by nonparametric color clustering

We present a color image segmentation scheme based on pixel color clustering. The proposed segmentation is based on mean shift clustering method, which is a powerful tool for the analysis of feature space. However, earlier clustering techniques based on mean shift use single scale over the entire feature space and are not feasible to analyze complex multi-modal feature spaces. In this paper, we present an adaptive mean shift method, in which the local scale information is involved. Actual regions in an image can be obtained by segmentation using the proposed clustering technique.