An algorithm of asteroid craters detection based on Gradient Vector Flow snake model

A method of craters detection based on Gradient Vector Flow (GVF) and Snake model is presented in this paper. Firstly, the original edge map of craters is detected by Canny operator, and the concave boundaries and ruptures are allowed in the edge map. Then, the external force of, which is called GVF, is computed as a diffusion of gradient vectors of the edge map derived from the image in the process of locating craters boundaries by the traditional active snake model. The problem of former poor convergence of concave boundaries can be solved. At last, the original positions of iteration processing of Snake model are judged by dynamic clustering algorithm, which improve the convergence speed. Experiments on real images of asteroid's surface including single crater and several craters are simulated. Results have shown that GVF-Snake model can locate not only the concave boundaries, but also the serious ruptures of edge image. The boundaries of craters can be detected wholly.