Research On Robust Video Watermarking

With the advent of the Information Age, information hiding technology has become one of important research subject in the field of information security. As a main branch of information hiding, digital watermarking concern on how to embedding the watermarking into the open digital media covertly without the digital media degradation. With the development of video compression technique, the application of video is more and more popular. Consequently, it is of great importance to study the technique of video watermarking.The main work includes the following respects:A video watermarking scheme robust to H.264compression. Aiming at the new video compression technology’s characters, a robust video watermarking algorithm based on block Discrete Cosine Transform (DCT)、Singular Value Decomposition (SVD) and Particle Swarm Optimization (PSO) is proposed. The stability of DC coefficient and singular values is utilized in the embed process. First, all the DC coefficients in luminance blocks are scrambled, and then SVD with block size of the scrambled DC coefficients is processed. The watermarking is embedded by modulating the third singular value. The PSO is used for selecting the appropriate embedding strength. Experimental results indicate that proposed algorithm makes a good tradeoff between invisibility and robustness compared with the algorithm with fixed embedding strength. In addition, this algorithm is robust against H.264compression.Sub-band switching based robust video watermarking for H.264video. One pair of predicted discrete cosine transform coefficients within the blocks of size4x4are used to embed1-bit watermark information. The embedding locations of DCT coefficients will switch from the lower sub-bands to the higher sub-bands, if the distortions of the predicted values are larger than a tolerable bound. This algorithm is robust against H.264compression. However, the robustness of this algorithm is discrepancy for the different types of video. In order to overcome the flaws, the block classification based on texture features is introduced. At the same time, the prediction functions for predicting the DCT coefficient is simplified. Experimental results indicate that the proposed algorithm enhances the robustness.