Adaptive active contours without edges for image segmentation

摘要

In this paper, pursuing the popular CV model for two-phase image segmentation and using Sobolev gradient, we develop a novel active contours model as an alternative solution to two-phase image segmentation problem. With our model, evolution velocity of level set function consists of adaptively-variable-sign velocity that is derived from the fidelity term of CV model and regularization velocity that is the Sobolev gradient of contour length. Because the first velocity has the opposite sign inside object and within background respectively and the same sign strictly inside object or within background, it can make the level set function move up or down according to image data. The proposed model has been applied to both simulated and real images with promising results.