Zero-watermark Research On Image

Internet provide people a way to communicate with each other and make our lifemore convenient, but in the same time it bring security problem of digital productsabout illegal copying and unauthorized modification issues, and all these matters wereextremely to be solved. In order to settle these matters, digital watermarking wassupposed and it provides an effective way for digital media copyright protection andthe blow of piracy, then copyright protection using digital watermarking technologyfor digital products have been recognized by the law. Development of digital imagewatermarking technology becomes more mature, they can already provide effectiveprotection for many digital products. But in face with some high-fidelity images, suchas medical, military and government documents and other digital medias, theembedded watermarking is restricted. in order to meet the needs of such products, thepaper propose following zero-watermarking algorithm for copyright protection.1.Introduce a zero watermarking method robust to print-scan attack. A method toenhance Harris corner is proposed in order to maintain high consistency. In addition,A new feature match is provided, which overcome the vulnerability of the originalfeature descriptor in face with print-scan attack. In watermark extraction, traversingsuccessful paring feature points, generating on-dimensional vector by using the localfeature information and the position message of the feature points. Then, with theSimilarity detection assessment value can determine whether the image to be detectedcontains watermarks. Finally, experiment show that the method provides robustwatermark information after a print-scan attack.2.Introduce an zero-watermarking algorithm based on the combination ofstructure tensor and LPT. By using the rich texture information if the image andgeometric invariant property of LPM, building a robust zeros-watermarking algorithmwho has the ability to resist geometric attacks. Firstly, the image is partitioned intoblocks, calculation the average of each block to get a low-scale image. Secondly, forthe low-scale image, using non-linear structure tensor extract a texture image and thenget the texture image’s LPM map by logarithmic polar transform. Finally, binarizingthe LPM map and performs an exclusive OR with logo watermark information to getregistered watermark to finish the copyright protection process. Experimental resultsshow that the algorithm has better ability to resist the conventional signal processing, the rotation, scaling attacks.3.Introduce two zero-watermarking algorithm based on the relationship offrequency domain parameters. Both DCT and DWT focus the image’s energy intofewer coefficients and have the same trend of the value changing in face with theattack. So comply with the trend of DCT and DWT coefficients, the paper proposedan algorithm to generate and zero-watermark information and XOR with owner’s logoto get and watermarking to apply for authorization. In addition, singular valuetransformation provide an more stable difference of the image transform coefficientsthe paper proposed another algorithm based on SVD,DCT and DWT. Experimentalresults show that these two algorithms, each one has their good point,and both ofthem figure out the problem of watermark transparency, robustness of contradictions,and common signal processing.