一种基于压缩感知的盲数字水印算法  

A BLIND DIGITAL WATERMARK ALGORITHM BASED ON COMPRESSIVE SENSING

作　　者：廖斌[1] 任美玲[1] 徐俊刚[2] 

机构地区：[1]华北电力大学电气与电子工程学院,北京102206 [2]中国科学院大学计算机与控制学院,北京100190

出　　处：《计算机应用与软件》2014年第2期307-311,共5页Computer Applications and Software

基　　金：国家科技支撑计划课题(2012BAH23B03)

摘　　要：现有的数字水印算法的研究成果大多在变换域,此类方法嵌入水印相对于空域而言数据量较大,安全性较高,鲁棒性较强,但是其计算复杂且抗攻击和防提取能力仍然有限。因此根据压缩感知CS(Compressive Sensing)的特点,结合人眼视觉系统(HVS)的特性,提出一种性能更高的数字水印算法,利用块不均匀度来选取容量大的子块,采用量化的方法自适应地选择量化步长来嵌入经过置乱处理后的数字水印。经实验验证,该算法保证数字水印具有更好的鲁棒性、隐蔽性和安全性,并且提高了含水印图像的抗攻击和防提取能力,而且提取水印不需要原始图像的参与,大大降低了存储载体数据的成本。Most research achievements of existing digital watermarking algorithms are in transform domain.Relative to the spatial domain, the watermarks embedded with these methods have larger data volume,higher security and stronger robustness,but it is complex in computation and limited in attack resistance and anti-extraction capabilities.Therefore,we propose a digital watermarking algorithm with higher performance according to the characteristic of compressive sensing （CS）and in combination with human eye visual system （HVS）.It selects the sub-block with larger capacity by the non-uniformity of the block,and embeds the scrambled digital watermark by using quantification method to adaptively choose the quantified steps.It is verified through experiments that the algorithm ensures the watermark to obtain better robustness,invisibility and security.At the same time,it also improves the capability of defending against attack and extraction of the image with watermark embedded.Furthermore,the watermark extraction does not need the involvement of the original image,this remarkably decreases the cost of storing carrier data.

关 键 词：压缩感知(CS) 数字水印 人类视觉系统(HVS) 量化 盲提取 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]