A content-based film video retrieval system named SupEye using MPEG-7(multimedia content description interface) was designed, which consisted of three parts: feature extraction, data storage and feature retrieval. The part of feature extraction took charge of the extraction of video features and video abstraction, in which assumptions and algorithms of video structure analysis and video abstraction were described on the basis of analysis on film video features. Data storage focused on the storage of video and feature data, in which multimedia content description files were divided into video features description and user interface description to improve retrieval performance through the usage of users′ retrieval features. The part of feature retrieval proposed a three-layered framework for multi-feature video retrieval which was composed of presentation, logic and transaction layer. On the basis of the analysis of MPEG-7, SupEye separates video features from user retrieval features and resolves the problems of feature analysis in sample query and the fusion of multi-feature query.
Dimitrova N, Zhang Hongjiang, Shahraray B, et al. Applications of video-content analysis and retrieval[J].IEEE MultiMedia.2002, 9 (3):42-55
Lienhart R. Comparison of automatic shot boundary detection algorithms Yeung, Minerva M, Yeo, et al. Proceedings of the SPIE. San Jose CA:Storage and Retrieval for Image and Video Databases VII, 1998:290-301
��ͨ,�ź꽭, ��ٸ�, ��. ��ͷ���ݷ�����������Ƶ�����е�Ӧ��[J]. ���ѧ��,2002,13(8):1577-1585 Lin Tong, Zhang Hongjiang, Feng Jufu, et al. Shot content analysis for video retrieval applications[J]. Journal of Software, 2002, 13(4):1577-1585 (in Chinese)
Rainer Lienhart, Silvia Pfeiffer, Wolfgang Effelsberg. Video abstracting[J]. Communications of the ACM, 1997, 40(6):55-62
Narasimha R, Savakis A, Rao R M, et al. Key frame extraction using MPEG-7 motion descriptors Matthews M B. Systems and Computers. California:Pacific Grove, 2003:1575-1579