ViBE A New Paradigm for Video DatabaseBrowsing and Search.pdfVIP

ViBE: A New Paradigm for Video Database Browsing and Search ∗†‡ Jau-Yuen Chen, C¨uneyt Tas¸kiran, Edward J. Delp and Charles A. Bouman Electronic Imaging Systems Laboratory (EISL) Video and Image Processing Laboratory ( VIPER) School of Electrical and Computer Engineering Purdue University West Lafayette, IN 47907-1285 Abstract shifted to key frame selection, shot clustering, and design of complete systems for extracting information from video se- Video Browsing Environment (ViBE) is a unique quences and presenting it compactly to the user [16]. Most browseable/searchable paradigm for organizing video of the work done has concentrated on processing a single databases containing a large number of sequences. The video sequence. This approach has the drawback of not ex- system first segments video sequences into shots by using ploiting shot clustering across sequences, which can sub- the Generalized Trace obtained from the DC-sequence of stantially limit the search power of the video database. An- the compressed data stream. Each video shot is then repre- other problem is that techniques designed to analyze single sented by a hierarchical tree structure of key frames, and sequences may fail to scale in performance when dealing the shots are automatically classified into predetermined with hundreds of hours of video of different types. pseudo-semantic classe