Shot Boundary Detection And Key Frame Extraction

With the rapid development of computer and multimedia technology, we have to deal with more and more video information. Facing so much information, it is very important to index the data we need easily and quickly. So Contented-based video Retrival(CBVR) becomes a hot research topic. The main tasks of content-based video retrival are shot boundary detection and key frame extraction.Firstly, the background and the development status of CBVR technology are introduced. Secondly, we synopsizes and analyzes the traditional methods of shot boundary detection and key frame extraction. Thirdly, we introduced the characters and color space in common use.Based on the methods of traditional shot boundary detection, one new method is put forward.Based on the differences of gray and color between the two frames, the detection for shot boundary and flashlight combines pixel-based method and histogram-based method. Experiment demonstrates it enable to locate abrupt boundaries exactly and quickly, but the effect of gradual boundaries detection has to be improvedAfter the analysis of the traditional methods of key frame extraction, a new method of key frame extraction based on cluster is puts forward. Every shot is divided into some mini shots, and then the frame having the biggest entropy will be extract from each mini shot as the key frame. This method overcomes the disadvantages of traditional method, which has fixed numbers and places of key frame.