Research of Image Segmentation Algorithm Based on Clustering

Abstract

Abstract. Through the study of existing image segmentation algorithms, this paper improves the standard FCM-based image segmentation algorithm. And according to the principles of effective treatment on neighborhood image noise by mean filtering and median filtering, proposes a new similarity distance calculation method, which gives full consideration to the gray information and neighborhood information of pixels. Experimental results show that this improved algorithm has higher segmentation efficiency, while having a high segmentation accuracy and strong noise immunity.