H.264 Watermarking Research Based On Chaotic Encryption

In recent years, as the intellectual property protection of multimedia information has been continuously strengthened in China, the application of video watermarking technology got improved at the same time. In a number of video compression standards, H.264/AVC for its extremely high compression ratios and good network affinity, quickly became one of the most popular video compression standards. Video watermarking based on H.264 becomes the focus of domestic experts and scholars to research. On the basis of studying and summarizing the achievements of H.264 video watermarking, some improvement ideas and schemes on two common problems, namely lower security and concealment due to not enough attention was paid on watermark image pre-processing, and video quality and watermark capacity cannot be both taken into account, was presented in this paper.This article introduces the emergence and development of digital watermarking, characteristic of video watermark and some key technologies in H.264 video coding standard. Especially made detailed description and analysis on encoding and decoding processes, intra and inter prediction. Then based on the compared the advantages and disadvantages of different areas where watermarks embedded in the H.264 encoding process, algorithm in this paper was determined which is embed the watermark in I frames by modifying quantized DCT coefficients.In respect of watermark pre-processing, characteristics and advantages of chaotic cipher in image encryption was introduced. Several common watermark preprocessing methods by chaotic system were listed and analyzed in area of security and real-time. According to the low security of Arnold and Logistic mapping, an improved method on encrypting and scrambling the watermark image was proposed. This new method improves the security by taking full advantage of chaotic sequences generated by higher dimensional chaos system and introducing more keys into encryption and scrambling process. The sequence just need to produces once during the entire pre-process to save time.In respect of Watermark embedding, texture-block-based embedding was introduced. By setting the number of nonzero coefficients of DCT matrix as threshold, different embedding methods were used on different blocks, which made the capacity effectively enlarged. Experiments show the video quality was guaranteed as well. In addition, the algorithm is robust against requantization attack and noise attack.