Texture analysis is an important research content in pattern recognition and computer vision, and we can get important information from the image through it. As an important method in feature extraction and classification, texture analysis has a very wide range of applications in the field of scientific research and engineering technology. In order to solve the problem of image classification, feature extraction and selection are combined in texture analysis. Different methods of texture analysis are used to extract texture features and a texture feature selection method based on grouped sorting is proposed in this paper. As in that work, 42 different texture feature parameters are extracted and efficient less ones are selected by the proposed method to be used in the classification. The proposed method is tested in numerical experiments on the texture database and real tumor data set. The experimental results suggest that the method is effective.