Research On Digital Watermarking Algorithm

Nowadays, technology develops rapidly;computer and Internet have becomean indispensable element in people's lives. More and more products exhibit topeople in digital form following by the increasingly flinty problems of copyrightbother. Digital watermarking technology as an important mean of copyrightmarking is playing a more and more important role in this field. Digitalwatermarking is a signal that is embedded into digital multimedia products inperceivable or unperceivable form for protecting copyright, verifying content, orproviding other information. The watermark integrated with the host informationclosely, and hidden in it, which has become an impartible member of the hostinformation, and could survive the manipulations which don't destroy the host'svalue of use and commerce. Digital watermarking technology relying on itsunique advantages can be applied in copyright protection,illegal copy,confidential communications, and many other fields, which has importantpractical significance and applicable value.This paper takes digital image as research object, has researched anddesigned embedding and extracting algorithms of the meaningful digitalwatermark.First of all, an algorithm based on DWT and fractal was proposed. After thehost information has been transformed by DWT, watermark information wasembedded while fractal coding part of the high-frequency coefficients. In order toreduce the time of fractal coding, as to any range block, we only found the mostsuited domain block in the higher resolution. The algorithm has syntheticallyconsidered the robustness and the quality loss of the image caused by the fractalcoding, the area with abundant texture was selected to be embedding points,which can reduce sensitivity of human's vision system to the descend of imagequality. In addition, the algorithm has also plotted out the search area of thedomain block more reasonably. Experiments showed that, while keeping thequality of host image, the algorithm kept the fractal code's preponderance ofrobustness. Because the algorithm decreased the time of embedding watermarkby the traditional fractal coding method, it enhanced the feasibility of fractalwatermarking method.Next, we made a quantificational analyses to the error caused by linearinterpolation applied in information hiding, base on which a splitting method wasproposed, and applied it in the watermarking filed, a robust digital watermarkingalgorithm based on DCT domain was proposed. The watermark information wasconfused with Arnold transformation and split to improve security and realizerecovered integrality. In order to improve the ability of resisting some attacks,and guarantee quality of the image, the split information was embedded intomid-low-frequency coefficients of the host image blocks which had beentransformed by DCT. Because only some mid-low-frequency coefficients werealtered, host image only changed a little. Embedding amount of the watermarkwas improved because of adopting the linear interpolation, and blind detectionwas realized. Experiment showed that, the algorithm is robust to JPEG lossycompression, cropping, noising, and so on.Again, a visible watermarking algorithm based on linear interpolation wasproposed. The sub image that would be hidden was transformed by DCT, in orderto compress the sub image, some DCT high-frequency coefficients were removed.The reserved coefficients were confused with a chaotic dynamical system toimprove the security. The confused coefficients were embedded intomid-low-frequency coefficients of watermark blocks which were transformed byDCT. Because the linear interpolation method was adopted, the problem that thehidden information is large of visible watermarking was resolved efficiently, andat the same time, the concealment, recoverability, and the quality of watermarkimage were guaranteed effectively. Experiment showed that, the algorithm isrobust to JPEG lossy compression, cropping, noising, and so on.Finally, a robust digital watermarking algorithm which is suitable forJPEG2000 was proposed. The watermark information was split and sorteddescending in the algorithm at first. At the same time the host information wasbroken into small blocks, which were transformed by DWT, and low-frequencycoefficients were quantized and sorted descending. Then the embedding points ofthe watermark were chosen among important quantization coefficients accordingto certain rules. In the course of extraction, if we do the same break, DWT andsort operation, we would accomplish the blind detection of the watermark.Experiment showed that, the algorithm is robust to JPEG2000 lossy compression.Additionally, because of the algorithm is block-based, it can also survive JPEGlossy compression.The algorithms proposed in this paper all aimed at meaningful digitalwatermark. Along with the development of technology and the progress of society,digital watermarking technology will also get rid of the stale and bring forth thefresh. I hope that the algorithms proposed in this paper could make slendercontribution to the development of digital watermarking technology.