Research On Image Segmentation Method Based On Texture Feature

Image segmentation has always been an important research topic in image processing ? computer vision and other fields. Texture segmentation is an important research method for image segmentation. Since most natural images have a particular texture characteristics, if not taken into account in the process of image segmentation, segmentation will lead to biased results or can not even get accurate segmentation results. Therefore,the image texture segmentation gets more and more attention from researchers. In this paper, for the problem of image segmentation influenced by texture,texture feature extraction and texture segmentation methods are adequately studied,the main contents are as follows:(1)For texture segmentation, complex texture region boundary description is not accurate, this paper proposes a method based on improved active contour model.The method extract the texture information from region boundary, it will be integrated into border texture information to the active profile energy functions, construct a texture-based energy as a function of external energy function, together with the role of the internal energy function,guiding the evolution of the initial contour to obtain the final target profile.Experimental results show that this method can effectively solve the problem of grain boundary area of the image description is not accurate,achieving the goal of obtaining a more complete outline.(2)For texture image segmentation region due to the texture features redundant and affect the efficiency of segmentation problem, this paper presents an improved K-means clustering segmentation method, by using principal component analysis of image texture features dimensional reduction in when using clustering method for dimension reduction texture segmentation features removes redundant information interference, improve the efficiency of segmentation. Finally, through experiments to verify the feasibility of the method.