A new digital watermarking algorithm based on NSCT and SVD

Abstract

Nonsubsampled Contourlet Transform(NSCT) take the advantages of multi-scale, multi-direction and translational invariance. In this paper, a new digital watermarking algorithm is proposed based on NSCT and Singular Value Decomposition(SVD). First, the low frequency component of image is got by apply Nonsubsampled Contourlet Transform to the original image. Second, the Singular Value Decomposition of low frequency component of image is taken. Last, the watermark information is embedded into the eigenvalue matrix of low frequency component of image. The result of experiment proves that the new algorithm can assure good imperceptibility for watermarked image, and the watermarks which is embedded using the algorithm have good robustness for familiar attacks, such as JPEG compression median filtering and added noise attack, especially good at anti-rotate attack, the NC of new algorithm can be increased by about 0.5 compared with watermarking algorithms based on DWT and SVD. © 2012 Springer-Verlag.