Abstract: The basic principles of curvelet transform are described and human visual model based on curvelet transform for optimizing the robustness of watermarking technology is given.In addition，the EMD algorithm with unique and efficient embedding approach which can be used to keep great balance between the robustness and transparency of watermarking is adopted for improving the capacity of watermarks.The proposed algorithm can not only effectively deal with all types of conventional signal processing attacks，but also can prevent most of the malicious attacks on watermarks.Furthermore，as a result of the characteristics of curvelet transform，extracting and detecting processes in this algorithm do not require participation of the original watermark and the original images.It is an utterly blind watermarking algorithm.Finally，experimental results prove that the algorithm is improved markedly compared to some other robust watermarking methods based on transform domains.

Key words: curvelet transform, human visual model, capacity of watermarking, robustness, Exploiting Modification Direction（EMD） algorithm

摘要：

阐述了曲波变换（Curvelet Transform）的基本原理，并提出采用基于曲波变换的人类视觉模型来优化数字水印的稳健性；还利用EMD（Exploiting Modification Direction）算法所特有的高效大容量嵌入方法来扩展水印的嵌入容量，使得数字水印在稳健性和透明性之间的平衡性上有了明显的提高。该算法不仅可以有效地应对各种常规的信号处理攻击，而且还可以防止大部分的恶意水印攻击。同时由于曲波变换的特点，算法的抽取和检测不需要原始水印和原始载体的参与，是一种全盲水印算法。最后的实验结果证明该算法相比于其他多种变换域类稳健水印算法，稳健性和透明性都有明显提高。

关键词: 曲波变换, 人类视觉模型, 水印容量, 稳健性, EMD算法