Content-based Video Retrieval Key Technologies

With the development of computer, internet and communication, the category and number of multimedia technology is increasing fast. It is an important task that how to find the information which we need from the large-scale database. Thus, Content-based video retrieval (CBVR for short), a new subject, comes into being.Together with MPEG-7, this paper tries to study some key techniques of CBVR model. The main researching content is about:Firstly, the content of video structure analysis and all kinds of algorisms was introduced in this paper. MPEG-7 descriptors and description scheme are detailed given. Also, we discuss the application of MPEG-7 descriptors and description scheme in organization features and partly carry out.Then, the paper tries for an actual problem of CBVR—image segmentation and brings forward a way of improving on k-mean cluster. A way of distinguishing texture pictures from lots of pictures was put forward besides.Thirdly, based on the content above, a universal video retrieval system is devised. Finally, the whole paper was summarized. The application prospect and trend of CBVR was approved.