Video model for dynamic objects

Content based retrieval from a video database depends greatly on how the semantics of video is described. In this paper we propose a video model that tries to capture the semantics of video segments based on the dynamic characteristics of the objects present in the segments. State-chart based formalism has been adopted to describe the objects of a video and their dynamic behavior. Several operators have been defined to operate upon the video segments. The video calculus introduced here helps in representing the semantics of the segments using first-order predicates. Formal semantics of this language has been analyzed and a simple implementation scheme for video databases has been proposed.