Citazione: A. Amato, M. Calabrese, V. Di Lecce, Relevance Feedback Oriented Cbir Interface For Semantic Discovery, Wseas Transactions on Computers, Issue 9, Vol. 5, pp. 1978-1985, September 2006. (ISSN: 1109-2750)
Abstract: Focus of this work is the presentation of a human-centric interface for a CBIR system. CBIR is oneof the most promising approach to the problem of the image retrieval. But it does notseem to be able to cover the semantic gap. In order to deal with this issue in the last generation image retrieval systems, user is seen as a part of the retrieval process andthis becomes a cycling one that allows the user to reach the desired results by means ofsuccessive approximations. The aim of the proposed interface is to show as much information as possible about the overall image database organization to user easily. User can interact with the system interface to give it relevance feed-back and to posemulti-image queries. The relevance feed-back mechanisms proposed in this work are user friendly and they allow user from different background to formulate complex queries andobtain good results without much training. From this point of view, the proposed interface can be seen as a tool allowing the semantic browsing of an image database.
Keyword: Images, Databases, Semantics, Retrieval, Queries, Approximation, Tools, Cycles, Training, Feedback, Browsing