A novel nonsampled contourlet-domain image watermarking using support vector regression

Abstract

Geometric distortion is known as one of the most difficult attacks to resist, for it can desynchronize the location of the watermark and hence causes incorrect watermark detection. It is challenging to design an image watermarking scheme robust against geometric distortions. On the basis of support vector regression (SVR) and the nonsubsampled contourlet transform (NSCT), we propose in this paper a new image watermarking algorithm with good visual quality and reasonable resistance to geometric distortions. Firstly, the geometrically invariant space is constructed by using image normalization, and a significant region is obtained from the normalized image by utilizing the invariant centroid theory. Then, the significant region is subsampled into four subimages. Finally, the NSCT of one subimage is obtained, and the digital watermark is embedded into the host image by modifying the low frequency NSCT coefficients. In watermark detection, on the basis of the high correlation among subimages, the digital watermark can be recovered by using the SVR technique. Experimental results show that the proposed image watermarking is not only invisible and robust against common image processing operations such as filtering, noise adding, and JPEG compression, but also robust against geometrical distortions.