Digital Watermark Algorithms Based On Characteristics Of Wavelet Coefficients

Along with the development of internet and multimedia techniques, more and more people pay attention to the security of multimedia information and its copyright protection, thus watermarking techniques appear. Whereas the maturity of the wavelet technology and its advantage in image processing field, a deep research of digital image watermarking technology based on wavelet analysis is done in this paper.Primarily, some typical image digital watermarking algorithms and their performance are analyzed. Wielding classification comparison and experimental methods etc., the superiority-inferiority of these typical image digital watermarking algorithms are summarized and the research thoughts of this paper are determined. and the key problems that need to be solved in this work are also concluded. On this basis, the statistical characteristics and their robustness which out of image wavelet coefficients are analyzed, and then the sensitivity of some characteristic information extracted on different attacks are analyzed by experimental methods, it provides the theory basis and experimental basis for follow-up work.Digital watermark Algorithm Base on Multiple Nonlinear Regression in Wavelet Domain is proposed. With this method, the original image is decomposed by 3-level-wavelet transform, and divide lowest approximation coefficient into N blocks with 3x3 sizes, then Multiple Nonlinear Regression model is established, which is employed to describe the statistics of Energy Value between center of block and neighborhood point. According to this Multiple Nonlinear Regression model, embedding algorithm and detection algorithm for watermark are designed. Simulation results show that the algorithm presented in this paper have good robustness and anti-attack ability in againsting JPEG compress attacks,image blighting attacks, median cut attacks, adding noise attacks and other attacks.A novel multiple semi-fragile watermarks cooperative authentication algorithm for image contents authentication is introduced. In this algorithm, multiple features are extracted from the image wavelet domain, and t watermarks are generated by extracting the image features. On this basis, a watermark embedding method based space geometric model is proposed, it can fight off effectively the effects of among multiple different watermarks. The incidental tampering location and classification of intentional content modification and incidental tampering can been completed by mutual cooperation among t watermarks, so it has provided an effective mechanism for image authentication. In addition, the watermark generation and watermark embedment are disposed in the image itself, and the information about original image or watermark is needless through the authentication based on content, as a result, it can enhance security and secrecy protection of watermark. Theoretical analysis and simulation results show that this algorithm provides a good authentication on content, and it is efficient in distinguish between intentional content modification and incidental tampering.