Research On Embedding Watermarking Algorithm Of Digital Images

As the development of the internet and multimedia, the transmission of digitalmedia has become more convenient. As the same time, the security of internet hasbeen more serious, especially the problem of the copyright of digital media. So, howto protect the copyright is urgent needed. This years, based on the traditionalencryption techniques, digital watermarking has caught people's eyes, anddeveloped quickly, with effective complementary techniques. The watermarking isthat embeds some digital information (auctorial digital ideograph, date, CO.trademark, etc.) into the digital production, to prove the copyright, the legal rightsand the interests of author.There are many methods to generate watermarks. In this article, we firstintroduce two watermarking based on Chaoses which are Logistic map and toralautomorphism map. Because the toral automorphism map is periodic, it is used forscrambling image usually, also for inv-scrambling to revert image. And, in theexperiment, we propose a new method for watermarking, which combines theLogistic map with the toral automorphisms. It is also proved to be effective in theexperiment.The watermarking schemes are also many. The article first just introduces astrategy called DCT block-based algorithm, which watermarks are embedded in DCcomponents of DCT domain. Discrete cosine transformation (DCT) is an orthogonaltransformation of digital image and signal processing, which can focus imageinformation and divide the image frequency into three bands: low, middle, high.And the mostly energy focuses on the DC component, so it has much largerperceptual capacity than any AC components. It is proved that the watermarkembedded into the DC component is more robust.We also propose a new scheme in DCT domain, making use of Watson'sperceptual model and incorporating its luminance masking with DC components.Watson's perceptual model is one of the best-known Human visual system(HVS) models. Watson Model estimates the perceptibility of changes in terms ofimage's block DCT, which is JND (Just Noticeable Difference). The schemeincorporating the Watson model with DCT block-based, classifies DCT blocks intothree classes, using luminance masking, and then embeds different intensitywatermarking into the every classes. Experimental results demonstrate that the embedding scheme is very robust.