Study on image segmentation based on dynamic clustering and real rough sets

The image segmentation method based on dynamic clustering and real rough sets (RS) was studied in this paper. First the appropriate clustering centers and numbers were found automatically by the normal dynamic clustering method instead of being pre-specified during image segmentation. Then the real rough sets theory was employed to separate each class into boundary region and lower approximation set. And the dynamic clustering method based on the generalized Euclidean distance was employed to segment the pixel's grey in boundary region in order to improve the segmentation effect. The experiment shows that this method can not only cluster the uncertain element in boundary region, but also segment the grey and color image. Additionally, the algorithm is simple and easy to be used and the segmentation effect of image is quite good.