ABSTRACT
Texture image segmentation plays an important role in computer vision based tasks such as object recognition and classification. The main idea in image segmentation is to obtain a partition of the input image into a finite number of disjoint homogeneous objects. Nevertheless, it is not always easy to get a good partition in regions with texture information, since it is difficult to differentiate the boundaries between two textures, due to the lack of sharp differences between them. Therefore, it is necessary to introduce extra features to the usual segmentation algorithm. Different approaches deal with the extraction of homogeneous features from textures.Manyof these texture descriptors go from statistical models to filtering methods, and to geometric approaches (Cremers et al. 2007).
