A robust dual color image blind watermarking scheme in the frequency domain

Aiming at the problems that many watermarking schemes cannot adapt to the image copyright protection with large amount of information, and the copyright protection of color images faces severe challenges, this paper proposes a robust dual color image blind watermarking scheme in the frequency domain. The scheme uses nonsubsampled contourlet transform (NSCT)’s good multi-directivity and multi-scale, translation invariance and Schur decomposition stability and anti-aggression to balance the invisibility and robustness of the watermarking scheme. Firstly, the host image is transformed by layered NSCT to obtain the layered low frequency coefficient matrix, and non-overlapping 4 × 4 coefficient blocks are selected in the layered low-frequency coefficient matrix, and the maximum eigenvalue of the coefficient block is obtained by Schur decomposition, and Quantized index modulation (QIM) is applied to the maximum eigenvalue to embed watermark information. When extracting watermarks, if watermarked image is attacked by geometry, and use image correction algorithm based on perspective transform to preprocess the image, which mainly includes feature point detection, image correction and sawtooth processing. The simulation results show that the proposed scheme has strong robustness to common signal processing attacks, and stronger and more effective robustness to common geometric attacks. Compared with the existing dual color image blind watermarking schemes, the proposed scheme has better performance.