一种基于PSNR的迭加量化公开水印算法  

An Iterative Quantization Public Watermarking Scheme Based on PSNR

作　　者：楼偶俊[1] 王相海[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院

出　　处：《中国图象图形学报》2007年第8期1333-1338,共6页Journal of Image and Graphics

基　　金：国家自然科学基金项目(60372071);辽宁省自然科学基金项目(2003205);辽宁省高等学校优秀人才支持计划项目(RC-04-11)

摘　　要：图像有好的视觉质量又要有高的鲁棒性是当今数字水印研究的热点与难点之一。提出了一种在确定图像质量的前提下具有高鲁棒性的公开水印算法,在其嵌入算法中,首先确定嵌入水印图像的质量——通过控制图像的峰值信噪比(PSNR)来实现,再根据图像所要达到的峰值信噪比来设计水印嵌入步长;另外,水印信号是通过权平均的方式迭加量化到图像DCT域的多个低中频系数上,根据人眼视觉系统(human visual system)的特性,按图像内容自适应量化嵌入,很好地解决了不可见性和鲁棒性之间的矛盾。实验结果表明,该水印算法能保证图像具有很高的视觉质量,同时对于常见的图像处理和攻击具有很好的鲁棒性。A good balance between image＇ s visual quality and its robustness is one of the hotspots and difficulties in digital watermarking. This paper proposes a public watermarking scheme which can achieve high robustness on the premise of keeping a predetermined visual quality. In the watermarking embedding algorithm, the quality of the embedded image is first predetermined by controlling image peak signal-to-nolse ratio （PSNR）, and the embedding step is then designed according to the target PSNR. One watermarking signal is adaptively embedded by means of weight average into a few iterative quantized low or middle frequency coefficients in terms of the characteristics of human visual system （ HVS ） and image content, so that a good compromise between the imperceptibility and robustness is reached. The simulation results show the proposed scheme can guarantee high visual quality of embedded image, and attain at the same time a good robustness to common image orocessing and attacks.

关 键 词：峰值信噪比 DCT 迭加量化 盲提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]