VIMS: A Video Information Management System

John Chung-Mong Lee, Qing Li, Wei Xiong

Abstract: Video data management is fast becoming one of the most importanttopics in multimedia databases. In this paper, we describe thedevelopment of an experimental video information management system,called “VIMS”, being implemented at the Hong Kong University ofScience & Technology, which employs two fundamental components—i)a Video Classification Component (VCC) for the generation of effectiveindices necessary for structuring the video data, and ii) a ConceptualClustering Mechanism (CCM) having extended object-oriented featuresand techniques. By incorporating CCM concepts and techniques togetherwith the classified features and indices generated from the VCC, theinformation management system enables users to form dynamically, among other things, video programs(or segments) from existing objects based on semantic features/indexterms. A prototype of this system has been constructed, using apersistent object storage manager (viz., EOS), on Sun4 workstations.