Research On The Key Video Mining Technology Of Content-based Compressed Domain

With the development of information techonology,the category and number of multimedia technology is increasing fast.However, the problems caused by visual information expansion are very serious, many fields make the acquired visual information idle as they can not effectively process them, which has caused a great waste of resources; in addition, the traditional video information management system that is once purely based on the text mode has also exposed obvious limitation. How to effectively organize, express, manage, query and retrieve video information has become an urgent demand,therefore,Content-based video mining(CBVM for short) is much concerned.Based on content video, the paper has carried out data mining in the video compression domain, and thus avoided the trifles of the previous data mining that needs a lot of prior knowledge, and the information system can be used once it is established, which has improved the accuracy of retrieval and reduced the redundant frame section of the video.In this paper,some research is introduced on the Video Mining in DCT compressed domain.The major research works of the paper are as follows:Firstly, the paper has particularly analyzed and discussed the basic framework of the content based video mining and made an analysis and discussion of the key tecnologies.Secondly, the paper has tried one of the current focuses in CBVM-shot segmentation technology and adopted an efficient correlation analysis-based segmentation algorithm:since the key frame extraction technology is another core technology of video retrieval,the paper has also adopted the rough set-based key frame extraction algorithm, which can effectively remove the redundant and improve the retrieval speed.Then, based on the above researches, the paper has also done corresponding simulation to the two algorithms mentioned above and got satisfactory results.Finally, the paper has briefly summarized the research contents of the paper and pointed out the research prospect and direction of video retrieval.