The Design And Implementation Of A News Video Search Engine Based On TV Slice

In this rapid developing information time, the magnanimous video data leaves people not knowing how to choose. Only depending on the artificial retrieving classification appears more and more tedious and low effects, thus the content based video retrieval becomes to receive people' s welcome. At present, the domestic various provincial broadcast television stations are the biggest content providers. Their websites own the copyright of these multimedia resources network. However, the organization of intermediary source used in the on-line direct broadcast, and the retrieval work primarily depend on manual processes, seriously losing the manpower, the physical resource and even lowering the efficiency.For this subject, this article has carried on the thorough study on theory and experiment.1, According to Jiangsu Broadcasting Corporation (JSBC) website' s special details, it used nowadays' world most advanced visual rhythm technology to transform JSBC website' s video on demand (VOD) system. Also, it designed and realized one automatic, highly effective, reliable news video search engine based on the TV slice. Aiming at the situation of the inaccurateness on locating to TV slice of single frame image, this article made the improvement to the algorithm of the visual rhythm, and designed the algorithm of visual rhythm to suit for the small portion of video comapre model. Thus, based on the single frame image compare, it realized the accurate location of the pieces.2, Simultaneously aiming at the characteristic of the advertisement program in the news, made some discussions and the research to the advertisement recognition work. Because of the characteristics of advertisement video, compared to the news video, include short lens, many shear, quick transform, it used the visual rhythm technology to obtain the advertisement position.The article finally provides the model system of the transformed JSBC website' s VOD.