Content-based description of images for retrieval in large databases: MUVIS

The rapid increase in the size of digital image and video collections is urging for the development of efficient browsing and search tools that skip the subjective task of keyword indexing, paving the way for the ambitious and challenging idea of content based description of imagery. With this goal in mind the MUVIS¹ system attempts to provide an integrated solution for the indexing and retrieval of images based on their content within large databases. In this paper we describe briefly the overall structure of the system and expose the promising results obtained so far, demonstrating the similarity retrieval capabilities of the system based on separate image features.