Supervised texture classification using wavelet transform

A multiresolution approach based on wavelet transform for texture classification has been proposed in this paper. The orthogonal and compactly supported wavelets are used to characterise texture images at multiple scales. The QMF bank is used as the wavelet transform to decompose the texture into sub-bands. The set of features, derived from the statistics based on first order distribution of gray levels, are then extracted from each sub-band image. It is shown that the multilayer perceptron with error back propagation algorithm increases the separability of features and gives better classification as compared to the minimum-distance classifier.