Robust Blind Image Watermarking Algorithm Based On Singular Value Quantization

In this paper, a new robust blind watermarking algorithm based on singular value quantization is proposed to solve the issues of image watermarking robustness, invisibility and blind extraction. In order to improve the security of watermark information, watermark image is firstly carried on the Arnold scrambling. Secondly, the original carrier image is divided into several blocks, and then, each sub-block is processed by RDWT. For each sub-block, the low frequency sub-band is taken out for further DCT, and the coefficients after DCT are decomposed by SVD. Finally, the watermark image is embedded into the first singular value of SVD, according to the quantization method. The experiment results show that the algorithm is strongly robust against attacks from JPEG compression, cropping, noise and so on. It not only can improve the invisibility and robustness of the watermark, but also can realize the blind extraction of the watermark successfully.