Multiple Video Watermarking Algorithm Research Based On DCT Domain

As the types and quantity of video products are increasing rapidly, it also brings an increasingly demand for copyright protection and content authentication. As a result, video watermarking technology is now widely used in these areas. Digital video is featured by large amount of data with excessive redundancy and non-uniform distribution of moving and nonmoving regions, which causes many urgent technical problems to video watermarking. Comparing with the static image watermarking, video watermarking is stricter in many aspects, e.g., higher security, lower complexity, higher robustness, blind extraction. In this thesis, a multiple video watermark(robust watermark and fragile watermark) based on original digital video is proposed by utilizing a preprocessing algorithm of watermarking image scrambling, discrete cosine transform(DCT), singular value decomposition and human visual masking characteristics.In order to improve the safety and robustness of video watermarking, a robust blind video watermarking algorithm based on original video is proposed. The human visual masking characteristics and DCT theory are combined in the algorithm. The scrambled binary image watermark is embedded in the DCT low-frequency coefficients of the original video frame’s green component. The experimental results show that the proposed algorithm has better imperceptibility, and are robust against attacks like the Gaussian,frame shear,Salt and Pepper,the compression, etc.In response to the large requirement of semi-fragile watermarking technology used in the content authentication, a semi-fragile video watermarking algorithm based on singular value decomposition and DCT is proposed in order to improve the robustness of the traditional fragile watermarking, and also the fragility when processing the conventional signal. The excellent characteristics of the singular value decomposition is utilized to strengthen the robustness of the semi-fragile watermark when processing conventional signal and improve the semi-fragile to malicious tampering of video content in the algorithm. A quantitative method is used to change the strength of the embedding location in the algorithm, which helps to achieve the blind extraction and detection of video watermark.Finally, the two algorithms are implemented in the same host video, and the multiple embedding and extraction of the robust watermark and fragile watermark in the same video carrier are achieved.