一种基于辐角的加性模型隐写分析算法  

The Steganalysis of Additive Modelable Information Hiding Based on the Arguments

作　　者：张敏情[1] 杨刚 吴旭光 

机构地区：[1]武警工程大学电子技术系,陕西西安710086

出　　处：《武汉大学学报（理学版）》2013年第6期545-550,共6页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然科学基金项目(61379152);陕西省自然科学基金基础研究项目(2012JM8014)资助

摘　　要：图像隐写分析理论研究越来越受到重视,分析特征的有效性的理论证明已成了隐写分析不可缺少的部分.本文根据图像在加性隐写模型下,隐写图像差分直方图傅里叶变换辐角等于载体图像差分直方图傅里叶变换辐角与秘密信息差分概率质量函数PMF(probability mass function)傅里叶变换辐角之和这一理论基础,提出一种新的隐写分析算法.根据隐写前后图像差分直方图傅里叶变换辐角的差异构造极大辐角矩最低有效位MSA(max sum of arguments)作为特征,从理论上对该特征的有效性进行证明,然后对典型的加性隐写算法LSB(least significant bit)替换算法和扩频图像隐写SSIS(spread spectrum image steganography)算法进行分析,在非压缩彩色图像库UCID(uncompressed colour image database)V2标准图库下进行仿真实验.实验结果表明,本算法对加性隐写模型具有很好的检测效果.Theoretical research of image steganalysis is being paid more and more attention on and the proof of characteristics is part of steganalysis now.Based on the theory that the DFT（discrete Fourier transform）arguments of the histogram of stegoimage is the sum of the DFT arguments of PMF（probability mass function）of stegonoise and the DFT arguments of the histogram of cover image,this paper presents a new steganalysis method.Firstly,the max sum of arguments（MSA）is created as the characteristics according to the difference of arguments between the DFT of cover image and stegoimage.Secondly,the characteristics is proved to be effective and then LSB（least significant bit）and SSIS（spread spectrum image steganography）is analyzed.Finally,our theory is verified with the experiments.The results of the experiment using the image library of UCID（uncompressed colour image database） V2 shows that this method has a good ability to detect the original images from stegoimages which are embedded with additive model.

关 键 词：隐写分析 差分直方图 离散傅里叶变换 极大辐角矩 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]