Mpeg4-based Video Watermarking Technology

As a main part of information hiding, digital watermarking is regarded as an effective solution to solve the security problem of multimedia protection. In this thesis, two typical watermarking systems for video content authentication and copyright protection are studied based on MPEG-4. Experiments are performed to test the effectiveness of proposed algorithms. The main research work and contributions of this thesis are summarized as follows:(1) The principle of digital watermarking technology, especially video watermarking is introduced in detail, including the origin and development, the classification and application, the standards for assessing, the classic algorithms and attacks of video watermarking based on the MPEG-4 encoding/decoding system. The compressing theory of XviD system is introduced briefly.(2) This paper proposes an improving Differential Energy Watermarking algorithm of adaptive embedding method based on video content character after plenty of research work on content authentication. The new algorithm consults the value of video image texture and content character. The adaptive embedding method ensures the watermark invisible and robust. Through experiments the method can hide watermark in MPEG-4 video stream more rationally, the capacity of watermark increases greatly and the authentication precision locates in block level.(3) Based on classic spread spectrum (SS) watermarking algorithm, a novel algorithm of multi-function watermark for dual authentication is proposed. Thanks to the excellent anti-jamming ability of spread spectrum and chaos sequence, the algorithm can increase the detection accuracy by replacing the coefficients of middle frequency in DCT domain. The dual watermarks are embedded in different region combined with MPEG-4 system in order to improve the security and the real-time processing capability.