Research In Content-Based Movie Video Retrieval And Automated Video Abstract System

Content-based video and audio retrieval system have been one of the research focuses in recent years. The so-called"content-based retrieval"corresponded with the past retrieval based on whole file is on the basis of the video data and correlates with the content. The video highlights belong to one kind of video abstraction, it composed of a series of wonderful clips abstracted from the original video.The film or movie is the most popular and common video resource. In this paper we use the film video as an example, aims to implement real-time retrieve of explosion and shot scenes, has carried on the following research mainly: first, we proposed a method of real-time recognition of explosions based on dominate-color and audio information and improved the accuracy the retrieve;second, we proposed a method of real-time retrieve of shots scenes based on audio information and differences between frames when the shot occurs;third,a whole video abstraction system is proposed, it can cut off the clips which have been retrieved and merge them into a independent video abstraction in the same time edit the information of the it ;forth, it offers two-level retrieval structure of"key frame-highlights"and can play segmentation result and the whole video just like a video player."Real-time Explosion and Shot Detection and Retrieval and Abstract System", which the arrangement is clearly demarcated, can successfully detect explosions and shots scenes in the film and satisfy the user's real-time retrieval. System goes on a large number of experiments and encloses the result of the test. The later research directions are also pointed out.