Abstract: A more secure image watermarking scheme is proposed using the visual cryptography in the pretreatment of watermarking.Firstly，the visual cryptography is utilized to hide the watermarking into certain pictures and one of encryptive pictures is chosen as final watermarking.Then，some suitable blocks selected by fuzzy cluster from the deoriginal image are made Contourlet transforming.The watermarking is embedded on the maximal singular values of the coefficients. The watermarking detection is achieved just by superposing the detected information and the certain pictures.The process needn’t a great number of operations and lots of cryptography knowledge.Experimental results show that the presented scheme also has high watermark robustness to clipping and common image processing operation such as noise adding，filtering，JPEG compressing.

摘要： 将可视密码学引入到水印的预处理中，提出了一种Contourlet域的安全性较高的图像水印算法。该算法首先根据可视密码理论将水印信息隐藏到一定数目的图片中，选取其中任一图片作为水印，然后对载体图像进行分块，通过聚类找出适合嵌入水印的块，并进行Contourlet变换，在变换系数的最大奇异值上嵌入水印。水印提取时只需用特定数目的图片对提取出的信息进行叠加，其过程无需大量计算和密码学知识。实验表明，该算法还具有抵抗噪声、滤波、JPEG压缩等常规攻击和抗剪裁等功能。