Abstract
Many multimedia applications, such as the World-Wide-Web (WWW), video-on-demand (VOD), and digital library, require strong support of a video database system. In this paper, we survey various approaches to content-based video data retrieval. Compared with traditional keyword-based data retrieval, content-based retrieval provides a flexible and powerful way to retrieve video data. We discuss three approaches in depth. First, since video segmentation is an elementary task in video index construction, we discuss an approach to detect shot changes for video segmentation, which is based on the processing of MPEG compressed video data. Second, since a video object consists of a sequence of frames, the techniques for retrieving video objects based on the similarity of key frames are addressed. Third, we discuss a video query model and video query processing techniques based on the symbols in a video object. The notion of two-dimensional strings is extended to obtain three-dimensional strings (3D-Strings) for representing the spatial and temporal relationships among the symbols in both a video object and a video query.