AN EXPERIMENT ON TEXTURE SEGMENTATION USING MODULATED WAVELETS

In this paper, a textured image is considered to be produced by a smoo th signal, representing in part the illumination environment, modulate d by some frequencies determined by surface characteristics of the tex tures. Through expansion of the smooth signal into wavelets, a texture d image may be decomposed into modulated ''wavelets'' providing multir esolution information. It is shown that the corresponding wavelet coef ficients can be obtained efficiently by using the standard wavelet tra nsform while the associated h and g filters are modulated accordingly. A set of multichannel filters is designed through the use of a modula ted ''wavelet'' multiresolution decomposition, providing both spatial frequency and orientation selectivity. Based on these multichannel amp litude responses as discriminating features, an image containing multi ple textures can be effectively segmented. The potential of this appro ach is shown by experimental results.