The Research And Application Of Texture Classification Algorithm

After decades of the research and development, and the classification of textureimage has made a lot of research results. The researchers proposed all kinds ofclassification algorithms of texture image. However, these algorithms only extract thetexture feature in the space or frequency domain, and there is a high computationalcomplexity, and low classification accuracy and poor robustness. Anyhow, so farthere has not been a great texture classification method, and the texture classificationproblem remains to be further research and exploration.In the existing algorithms, GLCM reflects the interdependencies of the grayscalein the space and some characteristics of the matrix can reflect space structure of thetexture very well. And wavelet packet transform is a very good tool that extracts theinformation of texture image in the frequency domain. It can not only do furtherdecomposition in the low frequency parts of the texture image, at the same time it canalso do further decomposition in the high frequency parts without omissions orredundancy. It is a good method of extracting features of the texture image infrequency domainTherefore, this paper proposes a texture classification algorithm based on GLCMand wavelet packet, and parameters of the model are optimized through crossvalidation. The experimental results show that compared to other algorithms thatextract texture features only in the space or frequency domain, this algorithm thatextracts texture features both in the space and frequency domain improvesclassification accuracy and enhances the robustness of the algorithm.