Classification of Remote Sensing Images using Wavelet Based Contourlet Transform and Accuracy Analysis of Classified Images

In remote sensing classification of spatial and spectral feature of multispectral images with high accuracy provide greater performance analysis. Wavelet transform is the most preferred transform for classification of both spectral and spatial features. The difficulty present here is the non-availability of directional features. The proposed wavelet based contourlet transform provide salient feature extraction using laplacian pyramid followed by directional filter banks. The extracted features were greatly reduced by using principle component analysis method. By that features the multispectral image has been classified into urban, wasteland, waterbody, hilly region by using fuzzy-c-means clustering algorithm. The wavelet transform is used for low frequency component classification and contourlet transform is used for high frequency component classification on the remote sensing images in the existing method. The aforesaid transforms provide less classification accuracy for remote sensing images. So it is proposed that the wavelet based contourlet transform is to be used for the analysis of both high frequency and low frequency component classification. Hence the proposed method shows that classification accuracy is higher than the existing method.