A Database Approach for Expressive Modeling and Efficient Querying of Visual Information

Abstract

With the rapid development of multimedia technologies, the number of available multimedia resources is always increasing and the need for efficient multimedia modeling, indexing and retrieval techniques is growing. In this work, we propose a hierarchical, hybrid and semistructured data model for representing video data. Based on this model, we build a declarative, rule-based, constraint query language enabling to infer and to retrieve spatial, temporal or semantic relationships from information represented in the model and to intentionally specify relationships among objects and events. We introduce the concept of temporal and spatial frame of reference which allows to simultaneously locate video contents according to multiple spatiotemporal environments in real world. Our model and query language are extensible, application independent, expressive enough and quite suitable for multimedia information retrieval.