Active Contour Model Based on Nonlocal Means Method For Color Texture Segmentation

摘要

Color texture segmentation is a very important subject in the fields of computer vision. In order to segment the color textures, a new method based on active contour model with nonlocal Tikhonov regularization is proposed. In detail, another smoothness item which is the Tikhonov regularization is added in the traditional active contour model. Meanwhile the nonlocal operator which is based on the image slice similarity is joined in the smoothness item. What's more, in order to improve the computation efficiency, this paper designs the Split-Bregman algorithm. At last, our performance is demonstrated by segmenting many complex and real texture images.