提出了一种图像内容可恢复的半脆弱水印技术，采用自适应水印对图像进行认证、定位篡改，再利用零水印技术从第三方处提取篡改处的恢复水印，重构被篡改作品。认证方案采用基于非抽样Contourlet域奇异值分解的自适应水印认证方法，该方法具有更高的篡改定位精度，能有效区分有意攻击与无意攻击。恢复水印是通过将原始图像分块、置乱后，分别进行4×4编码，再对码流加密后交给第三方获得的。实验结果表明，本算法能够在容许一定压缩比的情况下检测发生的篡改，并有效地恢复被篡改的区域。

In this paper, a new semi fragile watermarking scheme based on image content for tampering localization and recovery is proposed. The authentication watermark is embedded in the SVs (singular values) with an adaptive quantized method within the domain of nonsubsampled contourlet transform (NSCT). This method can improve the tampering localization capability without causing noticeable artifacts and can distinguish involuntary modification from vicious attacks. The recovery watermarking is achieved by dividing the original image, scrambling, encoding and encrypting, at last handing on the third party in which zero watermarking technique is used. The results of the experiment show that the proposed scheme can detect and localize the tamper, as well as can recover the tamper image when it undergoes JPEG compression.