Research On Compressed-domain Video Watermarking Techniques For AVS

With the development of communication technologies, the spread of digital multimedia information has become more efficient and extensive. However, the digital multimedia information is easy to be copied and modified. So its copyright protection and authentication becomes an urgent problem to solve. Digital watermark technology provides an effective solution. AVS was developed by China itself, and it has the international standards of video encoding with independent intellectual property rights. At present, the corresponding research of video watermarking is still in great lack. This paper, based on the summary of the existing digital video watermarking technology, tends to study the technology of video watermarking on the AVS compressed domain. The main work of the thesis is summarized as follows.(1) According to the security of watermarking and the problem of error propagation, a video watermarking algorithm based on chaotic encryption and error propagation control is proposed for AVS in compressed domain. This algorithm can improve the security of watermarking by modulating the watermark to chaos before embed. On embedding algorithm, it adopts two strategies for embedding of watermark in order to meet the different applications of visual effects and watermark capacity. Meanwhile, a compensation method to overcome the error propagation after watermark embedding into motion vectors, which will greatly decrease the error propagation.(2) Because the transmission environment is different, video is often compressed by different bit-rates. The video watermarking algorithm can also adapt to different compression bit-rates. A video watermarking algorithm based on Scalar Costa Scheme (SCS) and Watson model is proposed for AVS. Combined quantization parameters of video encoder and Watson visual model, this algorithm chooses the appropriate scaling factors and quantization parameters of the SCS watermark embedding algorithm automatically. This will enable embedded watermark algorithm to fit different compression rates by guaranteeing the security of the video quality. Experiments show that this algorithm has improved the robust of watermarking greatly on the basis of good video effect.(3) Nowadays, it is easy to tamper the video content by using multi-media software. For the authentication of video content, a content authentication watermarking scheme is proposed for AVS in compressed domain. It adopted the energy relationship between block of I frames, GOP structure information and frame structural information to structure signature. In the embedding algorithm,both of DCT coefficient and motion vector is treated to the carrier for watermark embedding. This algorithm on one hand enlarges the capacity of watermarking; on the other hand, it makes a comparison between the extracted dual watermarking and restructure of the signature to reduce the false alarm and missed alarm.