Digital Watermarking Research Based On Still Image

Digital watermarking technology in copyright protection, content authentication, copy control, information security and other fields are playing an increasingly important role. The rise of digital watermarking technology since the1990s, there has emerged a large number of research results. Nevertheless, in the field of digital watermarking there are still many problems awaiting solution, such as multi-watermarking technology, robustness, blind decteting, reversible, and openness, and so on.Paper focus on image watermarking algorithm based on transform domain. Against Discrete Cosine Transform domain and Discrete Wavelet Transform domain each propose a dual watermarking scheme. The algorithm idea of dual watermarking scheme in Discret Cosine Transform domain:take one low-frequency coefficient and a middle-band coefficient from the carrier image after Discrete Cosin Transform, embed watermark to low-frequency coefficient by using the multiplicative principle, embed watermark to middle-band coefficient by using the additive principle. The algorithm idea of dual watermarking scheme in Discrete Wavelet domain:embed watermark in the second decomposition horizontal detail subgraph and the third decomposition approximate subgraph by statistics-based approach, the statistical model calculated the average adjacent characteristics to each image pixel. Separately experiments for the two transform domain dual watermarking scheme were conducted, from the robustness point of view, Discrete Cosine Transform domain algorithm is better than spatial domain algorithm, Discrete Wavelet Transform domain algorithm is better than Discrete Cosine Transform domain algorithm.The paper also presents a novel digital image watermarking algorithm based on visual attention model. Extract visual attention regions and detect edge of the image, divide regions of the image, embed weak watermark information in the visual attention region and edge simple region, embed strong watermark information in the non-visual attention region and edge complex region. Experimental results demonstrate that this method can effectively reduce the visual distortion of the image, resist common image attacks, and improve the robustness of the watermark detection.