一种基于奇异值分解的稳健数字水印算法

提出了一种基于奇异值分解的数字水印算法，它对常见的失真是稳健的。由服从标准正态分布的随机数组成的水印被嵌入到图像分解后的奇异值中。实验结果表明本算法具有很好的稳健性。在经过了一般的信号处理操作和JPEG压缩以后，嵌入的水印能够被可靠地提取和检测。特别地，本算法对转置、行(列)镜像、直角旋转、行(列)整数倍放大和有填充平移等几何失真具有不变性。

A Robust Image Watermarking Algorithm Based on SVD

This paper proposes a watermarking algorithm based on singular value decomposition (SVD), which is robust to common distortions. The watermark, consisting of random numbers according to the standard normal distribution, is embedded into images singular values. Experimental results show thai the watermarking algorithm performs well in robustness. The embedded watermark can be detected reliably, following usual signal processing and JPEG compression. Specially, the watermarking algorithm is robust to geometric distortions, such as transpose, mirror reflection, right-angle rotation, scale, and translation.