Research On Video Watermark Algrithm For Synchronization Based On MPEG-4

With the rapid development of Internet and computer technology, people have entered into the digital information society. Information Era supplies many conveniences to create and transmit digital multimedia. At the same time, it also brings a severe problem-the piracy and distortion of digital multimedia. As an effective implement in copyright protection and integrity authentication for multimedia, digital watermarking has become an active and broad research topic. Synchronization attack is one of the challenges for blind watermark detection. If the detector's input is watermarked but synchronization fails, then the embedded watermark will not be detected. This paper mainly does some research on watermark synchronization technology, combining with MPEG-4 video compression standard.Three watermark algorithms are put forward in this thesis for MPEG-4 video:Firstly, MPEG-4 video watermark algorithm based on statistics of motion vectors. Taking into account the motion vector distribution characteristics of discrete,we divide groups by the amplitude of motion vectors, and embed watermark into the correspond groups bitwise. As the watermark is embedded repetitively and discretely, part of the video frame changes will not affect the correct watermark extraction.Secondly, MPEG-4 video watermark algorithm based on watson video model. In this method, watermark is coded with the technology of spread spectrum and embedded into the strongest sensitivity area of visual, DCT Luminance component. For better visual effect and embedded watermark information by maximum intensity, watermark capacity is control by Watson video model.Thirdly, video watermark algorithm based on DCT and motion vector combined. It improves the algorithm in the third chapter. For enlarging the watermark capacity, we divide the watermark and video into groups and embed the watermark into the corresponding video groups. Otherwise, it creates synchronization code for each group watermark and imbeds them into DCT coefficient of the corresponding video groups, and for improving ability of error coding, watermark is encoded by Golay code first.