Research And Implementation Of Advertising Video Copy Detection

The fusion of telecom network, internet and broadcasting network is a strategic goal on the development of Chinese information industry and an important direction. With the passing of the resolution by the State Council executive meeting at the beginning of2010, the process of the fusion of telecom network, internet and broadcasting network is accelerating, and the key part is the fusion of the internet and broadcasting network.As the internet and the high-speed broadband network technology become more and more robust, the internet has had the ability to send high-quality data, video and audio at the same time. Meanwhile, the TV users can have the unprecedented interaction way because of the bidirectional transmission mode of the internet after the fusion of telecom network and the internet.At the moment of the digital television coming into everybody’s homes, the new medium of television broadcast can increase users’ stickiness by deeply considering users’ demand. For example, removing the commercials of the TV programs will no doubt increase the users’ interaction and experience in the re-watching function of the IPTV. The commercials can be detected and segmented automatically, effectively and efficiently by the method of video copy detection. The existing copy detection algorithms mainly extract global feature to enhance the robustness and efficiency of copy detection. They can resist the general signal distortion well, but have difficulty in resisting some geometric attack, such as, rotating, scaling, clipping, etc. Aimed at the characteristics of exiting commercials in the TV programs, this paper puts forward a targeted copy detection method that can give consideration to the robustness, real time detection and the uniqueness so as to improve the recall ratio of advertising video measuring.