抗几何攻击的最低有效位数字水印算法

尽管最低有效位数字水印算法嵌入容量大,但抗几何攻击能力不足.通过分析几何攻击的原理,结合可视密码学和图像归一化技术,提出一种抗几何攻击的最低有效位数字水印算法.首先,利用可视密码学技术将水印隐藏到多幅图片中,同时仅选取一幅图片作为水印.然后对载体图像进行归一化,并以图像的几何中心作为中心提取重要区域.最后,依据重要区域的大小缩放水印,以增强水印与载体间的同步性,并从重要区域像素的低4位中随机选取嵌入位置通过异或运算嵌入水印.实验结果表明,算法具有良好的抗几何攻击能力.

Least Significant Bits Watermark Algorithm with Resisting Geometric Attacks

Although the least-significant-bits watermark algorithm has good embedding capacity, it has poor ability to resist geometric attacks. By analyzing the principle of geometric attacks, a new watermark algorithm with resisting geometric attacks is proposed by taking advantage of visual cryptography and image normalization. Firstly, the watermark is hidden into multiple pictures and one of them is chosen as final watermark. Secondly, the cover is normalized and the significant region is extracted from the normalized cover. Thirdly, the watermark needs to be scaled to fit the size of the significant region to enhance synchronization between them. And the lower 4 bits is selected from each pixel encapsulated by the significant region randomly and the watermark is embedded by performing XOR operation. Results show that the algorithm has a good performance in resisting geometric attacks. Key words: image processing; digital watermark; least significant bits; visual cryptography; image normalization