Multi-mode indices for effective image retrieval in multimedia systems

This paper presents a multi-mode indexing scheme for effective content-based image retrieval. Three types of indices are identified: visual indices for quantifiable visual information, semantic indices for non-quantifiable semantic information, keyword indices for keywords or free text. The underlying index structures are the HG-tree and the signature file. The HG-tree is one of the most promising multidimensional point index structures and the signature file is best known for handling keywords. The multi-mode indexing scheme combines and extends the HG-tree, the signature file, and the hashing technique to support a wide range of user queries in multimedia information systems. Experiments have been carried out on an image repository to demonstrate the effectiveness of the proposed content model and indexing scheme