The Research Of Watermarking Technology Based On H.264/AVC

With the rapid development of multimedia network technology, multimedia information, in particular the exchange of video information has reached unprecedented levels, followed by the video information security issues are also increasingly serious. Digital video watermarking technology is considered to the most effective methods to solve the problem of video information security. Digital video watermarking technology could be divided into compressed domain video watermarking technology and the original domain video watermarking technology by the carrier to be embedded. Usually, the video information is stored in a compressed format in the step of applications because of its great data volume. Consequently, the study of compressed domain video watermarking technology has a greater practical significance. As the video coding technology of H.264/AVC has better prospects, the paper select H.264 encoding standard, research watermarking technology based H.264 video .the main contents are as follows:Firstly, in-depth study the H.264 video encoding technology, which analyses the features of coding pattern, motion estimation, stream characteristics. Systematically introduce the basic theory of video watermarking technology and development status, summarize the advantages and disadvantages of the current watermarking based on H.264, and pose the new video watermarking scheme for H.264.Secondly,a method of watermarking suitable for copyright protection of H.264 is proposed. On the basis of the visual properties of human, the macroblocks with the coding models of intra₄×4 and more complex texture are selected to embed watermarking by the modulation of the DCT coefficients in mid-frequency. The intensity and location to be embedded are automatically decided by the bolck energy. The extraction of watermark is not in need of original video information and only needs decoding I frame.Thirdly, two watermarking schemes for video authentication suitable for H.264 are proposed. The first is integrity authentication which is based on the video bit-stream. In the compressed bit-stream of H.264, The authentication codeword are generated according to the mode of intra₄×4 in I_Slice firstly, and then embedded into B_Slice and P_Slice by modulating some VLC codeword of sub-block. This modulation is based on the mapping rule between the VLC codeword and the bit to be embedded. The mapped codeword is very similar to the original codeword, namely, the length of codeword is same and the value of coding element is similar. This scheme could detect the watermarking rapidly, which meets the requirement of the real-time processing of video. The second is suitable for the complete authentication of H.264/AVC video content. The proposed method first generated authentication code according to the encoding mode of macroblock, and then embedded them into motion vector by the modulation of the search points of motion estimation of 1/4 pixel. The step of modulating introducing the rate-distortion cost, the proposed scheme achieved a better balance of rate-distortion.Forthly, two information hiding schemes for video suitable for H.264 are proposed in the paper. The first is to use the prediction difference of intra₄×4 to embed the information. On the basis of the difference, for one thing the proposed algorithm divided the predicting modes of intra₄×4 (I4-modes) into two set in order to form the mapping rule between these models and the hidden bits, for another thing modulated these I4-modes in terms of the mapping rule to embed information. The hidden positions were chosen randomly by the Chaotic sequence that advanced the flexibility and security of the algorithm. The extraction of secret information is rapid in this algorithm, which fills the requirement of the video real-time processing. The second is to modulate the best predicting model to hide information. For macroblocks having the prediction model of intra 4×4, information hiding is implemented by adjusting the model of some 4×4 block.The location to be hidden is decided by the key. For other macroblocks in B or P frame, information is embedded by modifying the prediction model of the macroblock. An optimization handling for the inter macroblock is run to better the coding efficiency after the adjustment of pattern. The step of the modulation of mode introducing rate distortion, the proposed method got a better rate-distortion balance.