FEATURE INTEGRATION AND RELEVANCE FEEDBACK ANALYSIS IN IMAGE SIMILARITY EVALUATION

In this article we describe the results of a study on similarity evalu ation in image retrieval using color, object orientation, and relative position as content features, in a framework oriented to image reposi tories where the semantics of stored images are limited to a specific domain. The focus is not on a complete description of image content, w hich is supposed to be known to some extent, but on the extraction of simple and immediate features that can assure, through their combinati on, automated image analysis and efficient retrieval. Relevance feedba ck is introduced as an effective way to improve retrieval accuracy. A simple prototype system is also introduced that competes feature descr iptors and allows users to enter queries, browse the retrieved images, and refine the results through relevance feedback analysis. (C) 1998 SPIE and IS&T. [S1017-9909(98)00502-9].