Texture image segmentation algorithm based on Nonsubsampled Contourlet Transform and SVM

摘要

In this paper we propose a novel texture image segmentation algorithm base on Nonsubsampled Contourlet Transform and SVM. Texture feature of image is extracted through decomposing image which uses the characteristic of multi-scale and multi-directional of Nonsubsampled Contourlet transform, and then, classifying the Feature Image by K neighbor classification algorithm and training support vector machine. Finally, segment the whole feature image by means of support vector machine. Three synthetic textures image segmentation experiment and comparison with other segmentation method results show that the correct rate of the proposed method of texture image segmentation is over 98%, and the method can be good for texture segmentation.