In this paper, an active contour model is proposed for image segmentation, combining original image and feature image information with adaptive weight. The feature image is a texture descriptor which intrinsically defines the geometry of textures using semi-local image information. The Kullback-Leibler distance is used to design the energy of active contour model, in which the weight coefficient for original image and feature image is decided by the ratio of their distributions entropy. The energy minimization is achieved with Split-Bregman method. The results show that proposed method can ach...
