Research On Digital Video Watermarking Algorithms

With the development of network and multimedia technologies, digital media are often subject to illegal invasions such as malicious copy, delete, and modification. Copyright protection has become more and more serious. As a typical copyright protection technology, the digital watermarking technology has been extensively concerned.The digital watermark research mainly foused on image watermarking algorithm in the early stage. But With the popularization of digital video products such as VCD, DVD, as well as the rapid development of Internet video consumption, digital video watermarking has become one of the hot and difficult spot in digital watermarking research.Out of the current request of network video copyright protection and content management, this thesis investigates basic properties and extensive applications of information hiding technology, and presents two video watermarking schemes based on existing research results. Utilizing multiple signal processing technologies, these two schemes are typically aimed at two kinds of applications.MPEG-2 compression domain watermarking scheme for real-time video stream. Based on the MPEG-2 codec standards, this algorithm chooses the quantized DCT coefficients for watermark embedding. This thesis introduces the Sparsity of Matrix and Quantized Energy of AC sub-low frequency coefficients for the Q-DCT matrix. According to the Sparsity of Matrix, the algorithm embeds the watermark to the maximum Quantized Energy DCT block by modulating the AC sub-low frequency coefficients adaptively. Experiment shows that the algorithm has a lower computational complexity, better security, and the impact on video file size can be negligible.the Algorithm is real-time, can be easily combined with video codecs.Watermarking scheme for raw video stream. Based on the image watermarking algorithm, this thesis proposed and implemented a watermarking for video content protection in the view of video content protection. This algorithm chooses the key frames according to the intensity of NCG, devides the luminance block into sub-blocks, and then embedds the watermark into the largest singular value of the sub-block singular value matrix. This algorithm chooses different keys for different frames, and then uses the Kent function to generate the watermarking sequences. Each watermarking sequence is responsible for monitoring the corresponding frame; each bit of the sequence is responsible for monitoring the corresponding block. This scheme can detect whether the video frame content has been tampered and locate where has been tampered.Both of the watermarking algorithms can improve the video's visual quality, and hence better meet the performance requirements by different video service applications.