Abstract: The nonsubsampled Contourlet transform is a shift-invariant，multiscale，and multidirection transform，and has certain advantages in image processing.The watermarking algorithm is proposed based on the nonsubsampled Contourlet transform.The low-frequency coefficients concentrate the most energy of the image，embedding the watermarking in this area can increase the robustness.To meet the invisibility，it divides the low-frequency coefficients into 8×8 sub-blocks，adaptively embeds the watermarking according to the energy of each sub-blocks as well as human visual characteristic.Experimental results show that the algorithm has good robustness and invisibility，and effectively resists JPEG，noise attack.

Key words: digital watermarking, nonsubsampled Contourlet transform, shift-invariant, robustness

摘要： 非抽样Contourlet变换（NSCT）是一个平移不变、多尺度、多方向的变换，在图像处理中具有一定的优越性，提出一种基于非抽样Contourlet变换的数字水印算法，低频系数集中了图像的大部分能量，在低频嵌入水印可以提高鲁棒性，为了满足不可见性，对低频系数进行8×8的分块，根据每一块的能量以及人类视觉特性进行自适应的嵌入水印，实验结果表明算法具有很好的鲁棒性和不可见性，能够有效地抵抗JPEG、噪声等攻击。

关键词: 数字水印, 非抽样Contourlet变换, 平移不变, 鲁棒性