Automated Segmentation Using a Fast Implementation of the Chan-Vese Models

In this paper, a fast implementation method of Chan-Vese models is proposed, which does not require numerical solutions of PDEs. The advantages of traditional level set methods, such as automatic handling of topological changes, are also preserved. The whole process is described as follows: First, the Otsu thresholding method is adopted to obtain the initial contours for the following level set evolution. Then, the initial curves are evolved to approach the true boundaries of objects by using the proposed fast implementation method of Chan-Vese model. Experimental results on some real and synthetic images show that our proposed approach is capable of automatically segmenting images with a low time-consumption.