TEXTURE SEGMENTATION USING OVERCOMPLETE-WAVELET-FRAME BASED FRACTAL SIGNATURES

Abstract

This paper presents a variation of the wavlet-based fractal signature recently developed by Espinal, Huntsberger et al. Here, the overcomplete-wavelet-frame, instead of the standard wavelet basis, is used an multiscale signal/image representation to enhance the textural characterization by fractal signatures derived therefrom. Experimental results have demonstrated that this approach can improve texture representation and enable better detection of texture boundaries.