基于角点特征的几何同步数字水印算法

Geometrically synchronous watermarking algorithm based on the corner feature

对图像的角点检测与水印同步进行分析和讨论,在此基础上提出了一种基于角点特征的几何同步数字水印算法。为抵抗几何攻击,在归一化的图像中提取出较好的角点信息,重点分析了Harris与MIC角点检测的优缺点,将2种方法合并应用到角点检测,摒弃了单纯的像素信息修改,利用圆环的几何形状来统计表示水印信息。实验结果表明,所提出的水印算法较其他算法不仅能抵抗一般的图像攻击,而且在抵抗旋转、放缩等几何攻击方面的稳健性也有较好的提高。

Simple geometric attacks could destroy watermark or its position,causing the failure of most watermarking algorithm.A geometrically synchronous watermarking algorithm robust based on the corner feature was proposed.In this paper,the feature point was detected on the normalization of the image to resist geometric attack;the combination of Harris and MIC corner detected algorithm was utilized to extract steady corner based on analyzing the advantages and disadvantages of the two algorithm;the watermarking information was statistic embed in the shape of rings instead of changing the simply one pixel.Sufficient experiments demonstrate the robustness of this approach,especially in resisting geometry attack such as RST（rotation,scaling and translation）.