The Attack Model Watermarking Algorithm Based On Visual Characteristics Of Study

Digital watermarking is a new flourishing technique in information security, which achieved exponential growth both in research and application during the last few years. And because of the duality in security, it is called "the last line of defense".In this paper, we made an introduction of the status quo as well as a few remarks about the historical aspects of watermarking. By studying on different kinds of watermarking methods, we sum up the watermark algorithms and benchmark to provide a fair and uniform benchmark testing platform. Then we make a deep research on human visual system and wavelet transformation, and a new wavelet watermarking method based on HVS is proposed to solve the conflict between robustness and imperceptibility. Experimental results show high robustness of the scheme to most of the digital image processes such as JPEG compression, blurring, noise addition etc.Watermark attack which is taken as a most important part of watermarking system, is another key feature in this paper. In order to improve the performance of the watermarking system, we establish a linear attack model, and derive the rules of watermark attacks and defenses. A Power Spectrum Condition compliant watermark is then proposed to improve the HVS based wavelet watermarking method, which achieves superior robustness against linear attacks.