Region Segmentation of Color Images Using Self-Organizing Clustering Algorithm

Abstract

This paper proposes a new region segmentation scheme of the colored still image in order to apply to the structural description. At first, RGB image signal is mapped into HVC color space which is uniform and inherent to human perception. The region segmentation algorithm applies color difference defined in HVC as the scale. We establish the new algorithm of self-organizing clustering so as to get rid of conventional initial value problem. By the segmentation described above, it is clarified that our new proposed scheme is stable and perform a good segmentation.