Research On Compressed Video Watermarking Algorithm Based On Watson Visual Model

The digital media has become a faster and convenient way for information communication along with the popularization of Internet and the development of multimedia technique. At the same time, this also leads to serious problem regarding IPR protection. So security of multimedia information becomes an important issue. Digital Watermarking technology can embed copyright information into original data without any effects on use of the data, and the watermark can't be removed, only special extracting algorithm can extract it. Digital watermarking technology can identify the digital product which has been illegally copied, and it can protect the legal copied product. So digital watermarking as an effective method of copyright protection becomes an exciting research field, especially hot the video digital watermarking.This thesis focuses on video watermarking algorithms and principles, systematically summarizes the characteristics, the classification and the research status of video watermarking algorithms and principles, and focuses the research on the key techniques of video watermarking algorithms, which include watermark generation, embedding, detection, and extraction algorithms and watermark attacks. This thesis studies the differential energy watermarking (DEW) and the human visual system (HVS), which propose two kinds of video watermarking algorithm used in the DCT coefficient domain and VLC domain separately based on Watson visual model.This thesis proposes a kind of DCT coefficient domain differential energy watermarking algorithm based on Watson visual model by modulating the low-mid frequency DCT coefficients energy relationship to embed watermark. During the modulation, a theorem deduced from Watson visual model and energy modulation model is employed to restrict the modified magnitude of coefficients. Experimental results indicate that the presented scheme is transparent and robust to significant volumetric distortions (including additive noise, low-pass filtering and loss compression) and a part of frame attack (including frame shear, frame replacement).Then, This thesis also proposes a kind of VLC domain differential energy watermarking algorithm based on Watson visual model towards the impact to the previous algorithm by DCT quantify of video coding standard, which modulates the DCT coefficients relationship to embed watermark. During the modulation, a theorem deduced from Watson visual model and energy modulation model is employed to restrict the modified magnitude of coefficients. Experimental results indicate that the presented scheme is transparent and greatly simplifies the encoding and decoding complexity, in terms of real-time performance to achieve improved significantly.