Visual querying by color perceptive regions

A major research subject in image databases is to support efficient and effective access to images based on their visual content. In color image databases, this requires to support image retrieval by both global and local chromatic features. Image retrieval by color regions is complicated by the fact that regions which are requested in the query should correspond to relevant colored regions in the retrieved images. In the following we present the PICASSO system, which supports image indexing and retrieval based on colors. The system exploits a pyramidal representation of images. Multiple descriptions of image properties are created at different levels of resolution thus allowing effective retrieval through specific queries as well as imprecise ones. (C) 1998 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.