基于视觉密码和边缘检测的零水印算法

ZERO WATERMARKING ALGORITHM BASED ON VISUAL CRYPTOGRAPHY AND EDGE DETECTION

在版权认证领域,传统的水印算法往往是在载体图像中嵌入二值图像。这种算法有两个缺点:一是水印的嵌入改变了载体图像,而在医学和侦查等领域需要保留原始数据;二是图像受到篡改攻击后,很难判断攻击的类型和位置。针对以上两个问题,提出一种基于视觉密码和边缘检测的零水印算法。首先,对载体图像进行置乱和奇异值分解等操作提取特征矩阵;然后,将特征矩阵按照算法生成公有图和私有图;最后,将私有图和零水印存至认证中心。另外,应用游程编码对图像边缘进行无损压缩生成密钥,用来进行篡改检测。实验表明,该算法具有非常好的鲁棒性和安全性,尤其是抵御噪声攻击;在篡改检测方面也具有很好的效果。

In the field of copyright certification, traditional watermarking algorithm usually embeds the binary image into carrier image. Such algorithm has two shortcomings ： the one is that the embedded watermarking changes carrier image, but in the fields such as medicine and investigation there have higher demand in original data integrity; the other is that once the image is suffered from tampering attack, it is hard to judge the type and location of attacks. To solve the above two problems, this paper proposes a new zero watermarking algorithm which is based on visual cryptography and edge detection. First, it operates scrambling and singular value decomposition on carrier image to extract the characteristic matrix. Then, according to the algorithm it generates public graph and private graph from characteristics matrix. Finally, it saves the private graph and zero watermark to certification centre. In addition, it applies the run-length coding to make lossless compression on image edge to form secret key, which is used for tamper detection. Experiment shows that the algorithm has very good robustness and security, especially in noise attack resistance. It also achieves excellent effect in terms of tamper detection.