An architecture for content-based retrieval of remote sensing images

This paper presents an architecture of a system for content-based retrieval of remote sensing images (RSI). This architecture relies on combining results from image processing, pattern recognition and multimedia databases. RSI retrieval from a database is based on the notion of image similarity. A novel aspect in the paper is that the degree of similarity among images is refined during a query session, and is computed dynamically based on user-provided texture and color image patterns, as well as information on how images were pre-processed before being stored in the database