A comprehensive image similarity retrieval system that utilizes multiple feature vectors in high dimensional space

HyperMatch is a general purpose image retrieval engine made up of three components; a conventional database, a set of high dimensional image feature search indices and a coordinator. Various new and improved high dimensional indexing techniques are discussed for their applicability to different feature spaces and distance measurements. The role of the coordinator is to combine the search results from individual indices into an overall measure of similarity. Experiments reveal the coordinator must choose search parameters based on the number of features to be combined, data set size and the number of similar images requested. Important future development work includes index improvement and system evaluation.