Robust Watermarking Based on Norm Quantization Singular Value Decomposition and Zernike Moments

Many schemes for digital watermarking have suffered tremendous frustrations on geometrical attacks. In this paper, a new method is proposed to try to solve the difficulties. The original watermark is embedded into a standard image which is acquired through translation and scaling normalization to the host image. The standard image is divided into small blocks. After Singular Value Decomposition (SVD) to the block, the norm of block could be obtained. Then, the norm is quantified by a given step, and a watermark bit is embedded, the other bits are embedded similarly. In the extraction procedure, Zernike moments are used to estimate the rotation angle of image which may have been attacked. Then the image could be corrected with the angle. Therefore, the watermark can be extracted according to the embedding procedure. Tests show that the scheme in the paper is robust to compression attack and geometrical attacks, especially the rotation attack.