Adaptive Fast Image Segmentation Model Based on Local Feature

Region-based active contour model such as Chan-Vese model is able to handle the blurry boundary and complex topological structures in images segmentation. However, based on the intensity-homogeneous distribution, the effect on segmentation in the images with intensity inhomogeneity is not fine. Textures are fine scale-details, usually with some periodicity nature, and they cannot be detected by intensity information. Aiming at these problems, an adaptive fast image segmentation model based on local features is proposed. On the one hand, two kinds of region data terms are designed for detecting cartoon and texture parts respectively. The local statistic information is extracted in the adaptive patch to solve the over-segmentation induced by the intensity inhomogeneities. On the other hand, the texture feature information calculated in the adaptive patch acts to compute the Kullback-Leibler distance for detecting the texture part. Our model is solved by the split Bregman method for efficiency. Experiments are carried on both medical and texture images to compare our approach with some competitors, demonstrating the precision and efficiency of our the model.