A novel wavelet packet transform based feature extraction algorithm for image texture classification

A novel algorithm for extracting features of image textures to facilitate better classification is described. The algorithm uses energy matrices at all subbands of the wavelet packet transform coefficients to extract the feature vector. The size of feature vector is reduced by taking the weighted average of the feature vectors at all subbands. The classification is performed by Euclidean classifier. The experimental results are presented to demonstrate the effectiveness of the proposed algorithm. The classification rate of the proposed algorithm is found to be better than conventional wavelet packet signature (WPS) algorithm.