目的 提出基于NSCT-DWT-SVD结合的复合数字水印算法，实现水印信息的隐蔽嵌入，提高水印的鲁棒性。方法 首先对水印信息进行Arnold置乱，对载体图像进行NSCT变换，将产生的与原载体图像大小相同的低频子带进行小波变换，后对小波变换的低频部分进行SVD分解，确定嵌入强度后将水印信息叠加到奇异值矩阵中，实现水印的嵌入。结果 经仿真实验发现，该算法在保证水印嵌入信息量的前提下，满足了水印信息的隐蔽性，PSNR值达到30 dB以上，NC值达到了1，同时该算法对压缩、滤波、剪切攻击的鲁棒性较强，NC值均达到0.92以上，抵抗旋转攻击时仍能提取出较清晰的水印图像。结论 该算法满足了数字水印技术要求的隐蔽性和稳健性，并且可以对数字产品起到版权保护的作用。

This paper aims to propose a complex digital watermarking algorithm based on NSCT-DWT-SVD so as to achieve the concealed embedding of watermark information and improve the robustness of the watermark. First, Arnold scrambling was carried out for watermark information. NSCT was carried out for carrier images and then WT (wavelet transform) was conducted for the produced frequency sub-bands with the same size of original carrier image. Second, SVD decomposition was made for the low-frequency part after WT. After the strength of embedding was determined, watermark information was superimposed on the singular value matrix to achieve the watermark embedding. According to the simulation experiment, the algorithm met the concealment of watermark information while guaranteeing the amount of watermark information embedding. PSNR value reached over 30 dB and NC value at 1. Meanwhile, the algorithm had stronger robust against compression, filtering and cropping attack. NC value reached over 0.92. The algorithm still extracted clearer watermark images against spin attack. In conclusion, the algorithm meets the requirements of digital watermarking technology for the invisibility and robustness. It can play a role of copyright protection in the digital products.