一种最优意义下的数字水印算法

基于统计最优的意义下,提出了一种非对称鲁棒性数字图像水印算法.算法将水印嵌入到最少受图像操作的亚空间,使得嵌入的水印受到因图像操作带来的误差均方意义下最小;此外,通过对公钥的构造,从而实现误差无偏.同时,由于嵌入水印的矩阵(密钥)与提取水印的矩阵(公钥)不同,我们可以公开除了密钥外所有其它的信息.算法具有高的检测概率和安全性,低的虚警概率.通过各种攻击实验,表明了算法具有好的鲁棒性.

An Watermarking Algorithm Based on Optimization Statistics

Based on optimization statistics,we propose an asymmetric watermarking method,through which a watermark is hidden in the subspace that is least susceptible to potential modification,so that the watermark has Minimum mean square error(MMSE) followed after the image operation.In addition,we structure public key to realize the unbiased mean of error.Due to the difference between the embedding matrix(secret key) and extracted matrix(public key),we are able to release all information to the public,except the secret key.We can obtain a high detection probability and security,a low false alarm probability.The robustness of our method is demonstrated by various attacks through computer simulation.