基于内容的视频检索（Content based video retrieval）.doc

基于内容的视频检索（Content based video retrieval） Content based video retrieval Caoyaguang (Beijing Institute of Technology Institute of modern distance education, North Ship Institute, computer upgraded 2001, 2001023011076) Abstract: video retrieval has a wide range of applications and broad prospects for development, is the focus of multimedia research. The general implementation steps of content-based video retrieval are systematically introduced. From the analysis of system structure of content based video retrieval, comprehensive discussion of the system involved in the video structure, shot detection, key frame extraction technology, construct a scene from the video stream or group technology, feature extraction technology and video retrieval, browsing and retrieval feedback technology, analyzes the advantages and the realization method of various disadvantages of some key technologies in them, and introduces some new technology and methods. Finally, some questions for further study are proposed. Keywords: video database, content based, video retrieval, shot shear detection, key frame extraction, feature extraction, retrieval feedback 1 Preface With the rapid development of multimedia technology and network technology, video has been widely used in many fields. It is an urgent need to organize, express, manage, query and retrieve these large amounts of data that contain large amounts of unstructured information. Therefore, content-based video retrieval (Content-Based, Video, Retrieval, CBVR) has become a hot research topic in recent years. A content based video retrieval system is shown in figure 1. Fig. 1 block diagram of content based video retrieval system Firstly, the video structure is analyzed, and the video sequence is segmented into the lens and the key frame is selected in the lens. This is the basis and key to realize an efficient CBVR system. Then, the moving feature of the shot and the visual features in the key frame are extracted, and then stored into the