自嵌入完全盲检测顽健数字水印算法  

Self-embedding robust digital watermarking algorithm with perfectly blind detection

作　　者：叶天语[1] 

机构地区：[1]浙江工商大学信息与电子工程学院,浙江杭州310018

出　　处：《通信学报》2012年第10期7-15,共9页Journal on Communications

基　　金：浙江省教育厅科研基金资助项目(Y201017916)~~

摘　　要：针对现有顽健水印算法无法实现完全盲检测,将自嵌入思想引入到顽健水印领域,提出一种完全盲检测顽健水印算法。首先将原始图像分割成互不重叠的子块,对每个子块进行离散余弦变换,通过比较每个子块的直流系数与所有子块直流系数的均值之间的大小关系产生特征水印,利用Logistic混沌序列对特征水印进行加密,然后调整每个子块的2个离散余弦变换中低频系数的大小自嵌入加密的特征水印,最后进行逆离散余弦变换得到含水印图像。算法结合自嵌入加密的特征水印和盲提取认证水印实现完全盲检测。实验结果表明:算法在抵抗平滑、添加噪声、JPEG压缩、重采样、剪切和几何攻击如中间随机删除行、向下偏移行、向右偏移列上表现出很强的顽健性。Those existing robust watermarking algorithms couldn＇t achieve perfectly blind detection. A robust water- marking algorithm with perfectly blind detection was proposed by introducing the self-embedding idea into robust wa- termarking literature. At first, the original image was split into non-overlapping blocks, and each block was transformed by DCT. The feature watermark was derived from judging the numerical relationship between each block＇s DC coeffi- cient and average of DC coefficients form all blocks. After encrypted by Logistic chaos sequence, the feature watermark was self-embedded into each block by adjusting two middle or low frequency DCT coefficients. Finally, the watermarked image was obtained after IDCT. The proposed algorithm achieved perfectly blind detection by combining self-embedding the encrypted feature watermark and blindly extracting authentication watermark. Experimental results show that the proposed algorithm has strong robustness to resist various attacks such as smoothing, adding noise, JPEG compression, resampling, cropping and geometric attacks like random row removal, downward shifting and right shifting.

关 键 词：数字水印 自嵌入 完全盲检测 顽健性 特征水印 认证水印 

分 类 号：TN911.7[电子电信—通信与信息系统]