Research On Digital Watermarking Algorithm Based On Hash Vector Optimization

With the rapid development of the modern information technology and the Internet, various forms of multimedia works such as digital image, video and audio can be conveniently published and transmitted in the different forms via network. However, the problems of information security stand out seriously. During the transmission of information in the channel, it will often be affected by the attacker of his malicious tampering or intentional forgery. Thus, it will affect the decoders to decode the information and also the receivers to get the real information.Linear optimal watermarking algorithm based on hash vectors is studied, and new watermarking algorithms based on hash vectors optimization have been presented. The main contributions of this thesis are as follows.(1) The extraction and hesitation quantification of the hash vectors of images are studied.Firstly, the hash vectors of images were evaluated. Thus, the pseudo-random, robust and semi-global statistics of the images were extracted. Then, we carry out the hesitation quantification of the hash vectors, which is to complete the process of watermarking embedding.And a series of calculation is made on the watermark sequence. At last, we retrieve the watermark by establishing a linear optimization problem.(2) A new watermarking algorithm which combines the standardization of the hash values of the images with the linear optimization algorithm was proposed. Through the method of standardization processing, we are able to improve the visual quality of watermark image,enhance the randomness, the robustness of the algorithm and even the ability against the attack at the same time.(3) Another watermarking algorithm which combines the linear optimization of the hash vectors with the minimum of the watermark texture visibility was presented. By introducing the characteristics of the minimum of the watermark texture visibility, a new optimal watermarking algorithm was constructed. Experiments show that the visual quality of the watermark image was significantly improved and reduce the visual distortion of the images in the process of watermark embedding.(4) The distribution diagram of the Frobenius norm of watermarks were estimated, and semi-blind detection of the watermark was carried out. By making the design which is based on the specific attack algorithm, we present the estimated distribution result of the Frobenius norm of the watermark. Then we carry out the semi-blind detector respectively, aiming at resistingdifferent kinds of attacks. Experiments show that the distortion position and the distortion degree were detected more accurately.