The Key Technique Research Of Content-Based Video Retrieval

With the rapid development of the multimedia technology and network technology, and the maturity of modern computer technology especially data storage and transmission technologies, the video becomes indispensable information carrier day by day in such aspects as people's life , education , amusing and so on. Although video have powerfully expression ability,largely content of information,lifelike and vivid picture, but at the same time video also have shortcoming such as non- structural data format,enormous data amount and opaque displaying of the content, Let analysis and management of the video difficultly, such as browse and retrieval video. It is become an urgent demand that how to organize and manage these data effectively and how to store and get these date effectively by characteristic of multimedia data and how to find one's interested video segment easily from a great deal video data. This technology is Content-Based Video Retrieval that aroused attention generally. The content of the video includes information of the video structure,the vision and audition of the low-level and the semantic information of high-level, all of these are the foundation of edit and understand in the future. How to organic integrate all of these information, and stand up universal model of video's analysis and retrieval are still an extremely challenging subject of this domain at present.Only part of work on the Content-Based Video Retrieval are researched in this text, mainly including Content-Based Image Retrieval,Shot segmentation,Key frame extraction, and then organized a simple system of content-based video retrieval.To pick-up the lower-level information of the video's vision, we extracted the color and texture feature of video's Key frame, and then composed the color and texture feature to retrieval the Key frame; And for video structure information ,in this paper we have realized the segmentation of shot to construct the video, and the Shot Clustering is introduced only; Through expression the video shot by key frames, and extracted the key frames' lower-level visual information(color,texture feature) to complete the retrieval of video; The authors hierarchically cluster feature vectors in video database .The clusters here are called as index clusters .An index entry is created for an index cluster neighbors. The proposed method speeds up searching...