图像双层划分和奇异值分解的数字水印算法

为提高数字水印的安全性，提出了一个新的基于图像双层网格划分和奇异值分解的数字水印算法．讨论了图像双层网格划分的概念和方法，对基于奇异值分解的单层水印算法中的水印嵌入函数进行了改进．该算法采用传统算法对图像的第一层网格进行水印嵌入后，再采用具有补偿特性的水印嵌入方法，对图像的第二层网格进行水印嵌入，同时通过迭代过程，提取稳定的水印信息．实验结果表明，该算法对嵌入的水印具有很好的透明性，对重采样、平滑、加噪声和压缩攻击具有较强的稳健性.当一层水印被完全消除时，另一层水印仍能较好的被提取出来, 使数字水印的安全性得以提高．

Two-layer partition and singular value decomposition based image watermarking

For improving the security of watermark, a novel image digital watermarking algorithm based on two-layer gridding partition and singular value decomposition was proposed. The concept and method of two-layer gridding partition of an image were discussed, and then the watermark imbedding function of singular value decomposition based single layer watermarking algorithm was created. Watermark was embedded into the first layer partition of the image adopting traditional algorithm, then embedded into the second layer partition using compensatory watermark embedding method, and then stable watermark was extracted with iteration method, Experimental results show that the algorithm has a good transparence of the embedded watermark and is robust against attacks including resampling, smoothing, noise adding and compression. The algorithm can extract one layer watermark when another layer watermark is removed, and watermark security is improved.