A novel image segmentation method based on random walk

Image segmentation based on random walk model in graph theory can be transformed into large-scale sparse linear equations to solve problem. The final solution of the equation and the iteration convergence rate is depending on the selection of the initial value. It is a significant disadvantage for segmenting the large scale image while selecting initial value randomly. In this paper, we proposed a novel image segmentation method based on random walk model. First of all, we down-sampling the original large image to the small image which can be solved fast, then the small image segmentation leads to sparse linear equations of much smaller scale. After getting the solution, the probability results will be up-sampling to the up layer, and then solve the sparse linear equations in this layer; repeating this up-sampling process until to the top layer which is the original image. At last, segment the final probability image with a pre-set threshold. We test our algorithm on two natural images and compare the segmentation results with that from the original random walk algorithm. The segmentation results show that ours are much better. Our algorithm takes the low-scale image probability result as the initial value of the high-scale image segmentation process. Therefore, under the same calculation time, segmentation result by our algorithm is much better than that by the original random walk segmentation algorithm.