The Research Of Robust Image Watermarking Technology Against Desynchronization Attacks

Digital watermarking is an important branch of the information hiding, and is an effective means of copyright protection.Digital watermark is difficult to be removed and destroyed,which indicates for copyright protection.Digtial watermarking technology must have strong security,transparency,robustness. Three kinds of image watermarking algorithms are present, in some ways to resist the synchronization attack:1, Using normalization,construct the geometrically invariant space.Then, compute the wavelet moments with the normalized image. Select some low-level wavelet moments , the digital watermark are embedded into the original image,then quantize the magnitudes of the selected wavelet moments.2, Extract the feature points from the original image; then, obtain the local regions adaptively; finally, the digital watermark are embedded into ,then quantize the magnitudes of the wavelet moments.3, Using image normalization, subsample the significant region ,and decompose the two sub-images into the nonsubsampled contourlet transform domain respectively.Then embed the watermark information into the low-frequency region of optional one sub-image; When detection, use the SVR which is trained ,extract the digital watermark and the digital watermark can be correctly extracted under several different attacks.Simulation results show that the proposed watermarking schemes are not only robust against common signals processing, but also robust against the desynchronization attacks.