Abstract: In order to improve the performance of traditional random walk algorithm,an image segmentation algorithm based on toboggan and random walk is proposed.The image is segmented into a large number of small patches using toboggan algorithm on the basis of local gray value,and each of such patches is considered as a node,then the law of gravity is used to define weights between the nodes.The final segmented image is produced with random walk algorithm.The results show that our proposed algorithm possesses the nice properties of the toboggan algorithm and random walk algorithm,is both efficient and accurate.