A New Method for Grayscale Image Segmentation Based on Affinity Propagation Clustering Algorithm

The space complexity of Affinity Propagation (AP) clustering algorithm is high, and it is difficult to be applied to image processing problem directly. In order to overcome this shortcoming, this paper proposes a gray image segmentation method based on gray level histogram and AP clustering algorithm, named GLHAP. In the proposed algorithm, in order to decrease the input data of algorithm AP, only several high frequency gray values as the input data of the image are first obtained from the gray level histogram, and after processing the results of AP, image segmentation results can be obtained. Finally, the experiment results show that the proposed algorithm GLHAP is more effective and efficient than the compared with algorithms FCM and K-means.