一种基于HVS的小波域水印算法

A wavelet domain watermarking algorithm based on HVS

将有意义的二值图像作为水印图像,运用Arnold变换进行置乱加密,结合人眼视觉系统特性(HVS),对载体图像小波系数进行计算以得到自适应的量化步长,以该步长对水平子带HL和垂直子带LH的小波系数进行量化,并把量化误差平均分配到这两个系数上,在保证嵌入强度的同时减少图像失真。水印提取时不需要原始载体图像和水印。实验结果证明本文提出的水印算法具有较好的不可见性和鲁棒性。

Using Arnold transform to scramble watermarking image, calculate adaptive quantization steps from wavelet coefficients by using human visual system characteristic. Quantify HL and I.H wavelet coefficients with these steps and separate the quantization error to the two coefficients. Experimental results prove that this algorithm is robust enough to resist some image degradation processing such as clipping, JPEG compression, adding noise and so on.