Study And Implementation On Multiple Watermarks Embedding And Watermark Image Restoration

As the rapid development of internet technology, the illegal copying, forgery and distribution of digital product bring the huge loss to the author. How to carry out copyright protection and information security on the Internet has become one of the urgent problems to be solved in digital communication. This question has attracted much attention. The development of digital watermark technique has become a way to solve this problem.In this paper, the main task focuses on digital audio multi-watermarking scheme. In addition, a watermark image restoration scheme is provided at last:(1) In the first method, making full use of the very stable statistical characteristics of audio signal, the robust watermark is stored, in zero-watermarking form, in the key which is relevant to the statistical characteristics of the audio signal. In addition, in order to effectively reduce the bad effect of one-time large-scale shear operations, this paper proposes a patching-zero method in the shear location. In the process of watermark extraction, the original audio signal isn't required and the tampered region can be located very accurately.(2) In the second method, as extension to the color image as watermark still presents one of the open issues in watermarking research, a novel audio color multi-watermarking scheme based on LBG is proposed. Using the compression characteristics of LBG algorithm and the hidden characteristics of the linear instantaneous mixing model, the method find a balance between the transparence and robust.(3) At last, to restore the extracted watermark images which may be blurred due to the signal transfer or various signal processing operations, a watermark restoration system based on support vector regression is proposed. Using the structural risk minimization principle of SVR algorithm and the unique advantages of its training error and generalization ability, the method, combined with the mapping algorithm, built a model for each code vector, which classes the blurred sub-block image. Finally, the blurry watermark image is restored.Experimental results show that both of the two audio multi-watermarking schemes all can reach the requirements of transparency, robustness and security, and can be applied in copyright protection; the blurred watermarking image restoration scheme can improve the performance of the traditional watermarking system and achieve the purpose of the copyright protection more effectively.