ORTHOGONAL AND BIORTHOGONAL TEXTURE-MATCHED WAVELET FILTERBANKS FOR HIERARCHICAL TEXTURE ANALYSIS

This paper deals with the application of the wavelet transform to hier archical texture analysis. The main goal is the design of texture-matc hed FIR-filterbanks as the dyadic (2-channel) orthogonal and the M-cha nnel biorthogonal wavelet filterbanks. The new approach represents a g eneralization of the already known purely mathematically-oriented desi gn methods for multirate and wavelet filterbanks. The cascaded applica tion of the filterbanks results in a critically, by the factor M . M ( with M greater than or equal to 2) downsampled, pyramidal representati on of the textured image. The performance of these new filters discrim inating textures is compared with the classical wavelet approaches on Brodatz's textures.