MODELING AND CLASSIFYING SYMMETRIES USING A MULTISCALE OPPONENT-COLORREPRESENTATION

A new class of multiscale symmetry features provides a useful high-lev el representation for color texture. These symmetry features are defin ed within and between the bands of a color image using complex moments computed from the output of a bank of orientation and scale selective filters. We show that these features not only represent symmetry info rmation but are also invariant to rotation, scale, and illumination co nditions. The features computed between color bands are motivated by o pponent process mechanisms in human vision. Experimental results are p rovided to show the performance of this set of features for texture cl assification and retrieval.