抗几何攻击的多通道数字图像水印

A Novel Multi-Channel Image Watermarking Method for Resisting Effectively Geometric Attacks

文中提出一种多通道空间域抵抗几何攻击的数字水印方法,该方法首先将图像分解为多个通道图像,然后适当修改某通道图像数据并由其生成满足某种分布规律的水印模板,最后将该模板嵌入到图像的其它通道图像中。当图像遭受几何攻击时,由于生成水印的图像通道和埋植有水印的图像通道遭受同样的几何变换,因此算法实现了被埋植的水印和水印图像的自同步,回避了被嵌入水印和水印图像同步困难问题,从而具有较好的抵抗几何攻击能力。文中通过实验,验证了算法不但可以抵抗仿射变换攻击,而且可以抵抗多种几何变形攻击。

Aim. At the beginning of this paper, the review of existing literature on resisting geometric attacks^[1-8] reveals, in our opinion, that none of them is sufficiently effective. We now present a method that, we believe, is quiet effective. In the full paper, we explain our novel method in some detail; in this abstract, we just add some pertinent remarks to listing the two topics of explanation. The first topic is. the principles of multi-channel image watermarking framework. Its two subtopics are. the generation of the image watermarking template from image channel＇s data to be embedded （subtopic 1.1） and the embedding of image watermarking template （subtopic 1.2）. In subtopic 1.1, we make use of the chaotic watermarks proposed in Ref. 9 by G Cotsman et al. In subtopic 1.2, eq. （4） in the full paper is taken from Ref. 10. The second topic is. the detection of image watermarking. In the second topic, what is done in topic 1 ensures self-sychronization so that image watermarking can always be detected effectively by our method not only for affine transmission, which can be detected by existing methods though not effectively, but also for other cases of geometric attack, which can not be detected at all by existing methods. Experimental results, given in Figs. 3 and 4 in the full paper, demonstrate preliminarily that this image watermarking produced by the multi-channel framework achieves a high performance of robustness even if an image undergoes some serious geometric distortion attacks.