Image retrieval based on chromaticity moments
George Paschos, I. Radev · 2003
A number of different approaches have been presented in the recent past addressing the issue of image retrieval using color and texture features. Most of these methods are based on elaborate mathematical modeling, including auto/cross-correlation, random fields, and Gabor multi-channel filtering. Furthermore, the computational requirements of these approaches are usually high. A much simpler yet effective method for color texture retrieval is presented here. The method uses the chromaticity diagram (x, y) and a corresponding set of two-dimensional and three-dimensional moments to characterize a given color texture. Chromaticity is thus described by its 2-D shape on the x-y coordinate system and the associated 3-D histogram. Using a relatively small number of moments the method achieves retrieval rates of over 90%. The method has been applied to the retrieval of granite and marble images.