几何攻击会给数字水印带来同步误差和插值误差,现有的多数抗几何攻击的鲁棒水印方法都把焦点集中在同步误差上,对插值误差的研究甚少。介绍了常见的插值算法并指明了插值误差产生的原因,能够实现最佳的嵌入率、嵌入失真以及鲁棒性之间平衡的QIM(quantization index modulation)水印算法,进而分析了插值误差对QIM算法的影响,在此基础上提出了针对插值误差的逐像素点选择QIM水印算法。实验在纹理程度不同的10幅图像上进行。实验证明,提出的水印算法对插值误差的鲁棒性优于原始的QIM水印算法。
Geometric attacks bring synchronization and interpolation errors to digital watermark. Existing robust watermarking schemes focus on synchronization errors, but fewer papers pay attention to interpolation. The quantization index modulation (QIM) method was presented which achieved very efficient tradeoffs among the amount of embedded information, the amount of embedding-induced distortion to the host signal, and the robustness to intentional and unintentional attacks. The influence of interpolation errors on QIM method was analyzed. Pixel-selection-based QIM robust watermarking scheme was presented. Experiments results on 10 standard images show that the scheme outperforms a better robustness against interpolation errors than QIM method.