Research And Implementation Of Video Summarization Technology

Digital video belongs to the field of information industry which the Country develops especially. With the rapid progress of broadband network and digital TV, the applications of video-on-demand, interactive TV and video website become more and more popular. All of these applications are facing the emerging digitalized video data in a large amount. But as we know, video information is linear and non-structural, which makes video browsing to be time-consuming and boring. Video summarization, aiming at reducing redundant data and improving browsing efficiency, is very important in various video-based applications.In this paper, we roundly conclude the conceptions , characteristic needs and implementation strategies of video summarization. After a survey of the research work in the literature , this paper makes a conclusion of the normal process of video summarization. Novel algorithms for generating static and dynamic video abstracts automatically are presented.In this paper, a quick sub-shot segmentation algorithm is developed to perform the video segmentation task. A keyframe extraction algorithm explores the results. It also makes use of a series of target such as color abundance, faintness and face amount to support selecting of keyframes and a better representation of static summarization.We investigate using multi-media fusion approach to resolve the problem of film content extraction and presentation. We present an approach of event-oriented feature abstracting. An algorithm for scene significance estimation integrating impressive and exciting extent, text and face appearance, audio feature, shot duration and other facets is presented.We find out that almost all the schemes pay much attention to how to get the summarization and always ignore the summarization model itself. Furthermore, it is so hard to communicate and share information between those schemes. Consequently, we do research on video content description and utility calculation based on the "Entity-Description-Utility" model.There are novel ideas in our approach on video abstracting. We define a measure of visual complexity of a shot, and map complexity to the minimum time for comprehending the shot. Then, we employ syntax rules for scene removal. The target skim is created using a generalconstrained utility maximization procedure that maximizes the information content and the coherence of the resulting skim.Finally, the work background and application are introduced, the conclusions are made, and some promising research directions are given.