Multi-region Segmentation of CT Images Based on Information Fusion
Li Jin, Zhou Lulu, Hong Zhi Yu, Liang Hong · 2006
In order to make both the edges and the multi-regions of a CT image more clearly, a new multi-region segmentation method based on information fusion, in which clustering segmentation algorithm is combined with edge detection, is proposed in this paper. Firstly the CT image is segmented by the K-means clustering algorithm and Canny edge detection operator to obtain two images with clear regions and distinct edges respectively. Secondly these two images are synthesized to form a new one by the information fusion method based on features. The advantage of the method is that value K and the initial clustering centroids are determined automatically according to the image histogram and no need for the rectification of images before the images are fused. Experimental results show that the images are clearly segmented into multi-regions and the distinct image edges are preserved simultaneously as well