Fast Texture Segmentation Based on Semi-Local Region Descriptor and Active Contour

In this paper, we present an efficient approach for unsupervised segmenta- tion of natural and textural images based on the extraction o fi mage features and a fast active contour segmentation model. We address the problem of textures where neither the gray-level information nor the boundary information is adequate for object extraction. This is often the case of natural images compose do f both homogeneous and textured regions. Because these images cannot be in generaldirectly processed by the gray-level information, we propose a new texture descripto rw hich intrinsically def ines the geometry of textures using semi-local image informatio na nd tools from differen- tial geometry. Then, we use the popular Kullback-Leibler distance to design an active contour model which distinguishes the background and textures of interest. The exis- tence of a minimizing solution to the proposed segmentationmodel is proven. Finally, at exture segmentation algorithm based on the Split-Bregma nm ethod is introduced to extract meaningful objects in a fast way. Promising synthetic and real-world results for gray-scale and color images are presented.