Texture Image Segmentation And Analysis Based On Schrodinger Equation

With statistical significant similarity or repetitive of the structure, texture shows a relatively universal basic property of the surface in nature.Therefore, according to these characteristics, the texture image segmentation divides image into different areas. However, because of the various natural texture types, different forms and complicate structures, texture segmentation is still challenging in the field of computer vision researches. Researchers have given many kinds of texture image segmentation algorithms in the long phase research, such as FCM, markov random field and base on filter bank. Although these methods have improved the results of image segmentation in some extent, understanding the human vision system is different, base on the multiple natural texture types, different forms and complicate structures. The segmentation of textured image is a major problem in image processing.In recent years, researchers have found some texture image segmentation methods based on the simulation analysis of visual perception, for example Gabor and multi-scale wavelet analysis. However, these methods carried out a large amount of computation. In order to effectively analyze the texture image, this paper carried out a texture segmentation method based on Schrodinger equation and improved FCM, it testified that the method can improve the result of image segmentation.Firstly, the paper proposes texture segmentation based on Schrodinger Equation. Texture image is the wave function of Schrodinger Equation in the method, and gets the potential energy function in the texture image processing through Schrodinger equation.Then, base on texture analysis of the schr?dinger equation, the method describes a method of fuzzy C-means and the coupled hidden markov random field. The method extracts the image feature using the label field of texture image through FCM and texture image parameter estimation of markov random field, then gets the result of image segmentation.At last, we compare our approach with other classic algorithms in segmentation of synthetic and brodatz texture mosaics. Large numbers of data indicate that the method of this paper improves the result of texture image segmentation, especially structural texture segmentation. We ensure that this study will help the development of texture image analysis and segmentation theory and method.