Color image segmentation algorithm: An approach to image segmentation through ellipsoidal clustering and edge detection

The proposed algorithm approaches color image segmentation though the detection of edges on regions of homogeneous color. Ellipsoidal clustering is performed on the RGB color space through orthogonal regression. Edge detection is performed using a novel non-linear filter resembling a second central moment. Detected edges are binarized resulting in closed regions suitable for region filling. Subsequent region merging produces the final segmented image. The primary contributions of this research are the first two steps of ellipsoidal clustering and color edge detection.