为了实现数字图像的版权保护,基于双树复小波变换,提出了一种新的零水印算法.该算法由于不改变图像的任何信息,因此在兼具不可见性和鲁棒性的同时,还可以解决常规的冗余变换域水印算法的能量损失问题.该算法先借用实小波变换的零树结构思想,在变换后图像中选择重要的系数树,并利用主分量分析提取它们的第一主分量,然后经过量化编码构造零水印信息,再到认证中心注册后,即可作为用户的版权标志.实验结果表明,该算法不仅具有很好的鲁棒性,而且可以抵抗滤波、加噪、有损压缩等各种攻击.

To achieve the copyright protection of digital image,we present a novel zero-watermarking algorithm based on DT-CWT(dual-tree complex wavelet transform).The algorithm doesn't modify any information of the image,therefore it possesses both the invisibility and robustness,while it can resolve the loss of energy of conventional watermarking algorithms based on redundant transform.Exploiting the idea of real wavelet zerotree structure,we choose the vital-coefficient-trees from the transformed image,then by using PCA(principal component analysis),extract the first principal components,quantize and encode them to construct the zero-watermark signal,which is registered to the center of authentication and is used to identify user's copyright later.Experiments show,that the robustness of the algorithm is very good,and it can resist various attacks such as filtering,noising,lossy compression,etc..