Texture-adaptive mother wavelet selection for texture analysis

We discuss the use of texture-adaptive mother wavelets in an adaptive probabilistic wavelet packet approach to texture analysis. First, we present the use of adaptive biorthogonal wavelet packet bases in such ananalysis, thus combining the advantages of biorthogonal wavelets (FIR,linearphase) with those of a coherent texture model. In this case, the computation of the probability uses both the primal and dual coefficient of the adapted biorthogonal wavelet packet basis. The computation of the biorthogonal wavelet packet coefficient is done using a lifting scheme, which is very efficien in terms of reducing the computational complexity and achieving an intrinsic parameterization of wavelet filters Then we include the mother wavelet parameter into this model, in order to fin the optimal mother wavelet for a given texture using this model. The model is applied to the classificatio of mosaics of Brodatz textures, the results showing improvement over the performance of the corresponding orthogonal wavelets