In this paper ,the method of texture information extraction based on wavelet transform is discussed in detail,then the Principle Component Analysis(PCA) is performed to SPOT5 image,using software MATLAB , the texture information extraction of the first principle component basing on wavelet transform is completed ,and two type of texture information are achieved ,they are mean texture image and standard deviation texture image.Through analysis of the texture information and band combinations,he selected bands with texture information are used in classification,the method of Maximum Likelihood Classification is executed in the study,comparing to the applying the spectral information only,the overall accuracy raises from 67.3% to 81.9%, and the kappa index raises from 0.676 to 0.764 when using texture information.The result showed that classification overall accuracy and kappa index are all improved in the study .
展开▼