Semantic Object Based Retrieval from Surveillance Videos

In recent years, due to technological developments, the use of Closed-Circuit Television monitoring has been widely used not only in public areas but also in confined and/or private spaces for improved personal safety and security. The increased data acquisition has naturally resulted in the critical need for multimedia analysis for semantic object and event detection. Addressing this research problem, in this paper we present an novel architecture for extracting and indexing semantic objects with Scale Invariant Feature Transform features. The proposed approach exploits the developments of motion tracking and video indexing algorithms. The proposed framework is an ongoing development with the objective to enable the semantic retrieval of objects. The preliminary performance analysis of the proposed approach has been evaluated on a set of surveillance videos.