The Research Of Texture Image Segmentation

The most important step of image processing is the image segmentation, because the results of image segmentation can directly affect the precision of followed procedures. The texture is one of the most important features to images, it can be found in every image. Because of the complexity of texture, the segmentation of texture image is especially difficult, and it seriously restricts the development of image processing.Research was made on the texture image segmentation algorithm, the processing have two parts, one is the texture features distill based on texture feature distilling algorithm, The other one is get segmentation result by classifying texture features. The main work is focused on these aspects.1. Systematically studied the characteristics of mathematical algorithms of texture image description, the texture feature extraction algorithm, as well as image segmentation method based on texture features. After reading of a lot of disquisition and experiments, gray level co-occurrence matrix were used to extract the texture features of images, after that experiments were made,and the results of experiments show that the use of Gray Level Co-occurrence matrix texture features extraction can get better segmentation results;2. Made full researches on fuzzy clustering algorithm, especially the fuzzy C–mean clustering. Proposed a new C–mean clustering algorithm. This algorithm uses modified distance function, data pre-processing and cluster center initialization using Hard C-means clustering algorithm, in order to improve the performance and reduce the clustering time.3. Studied RBF and RBF possibility artificial neural network, by comparing, RBF possibility neural network was chose, depicted the theory of RBF possibility neural network, and presented the network building, parameter setting and network training method. By experiments, the results show that the using of RBF possibility neural network can receive impressive image segmentation results.The algorithm is proposed to segment the texture images in texture image library, but the essential theory is also the same with other types of images. It can be conveniently applied in other image by modifying a small quantity of parameters.