基于图像高阶MARKOV链模型的扩频隐写分析  

Steganalysis of Spread Spectrum Image Steganography Based on High-Order Markov Chain Model

作　　者：张湛[1,2] 刘光杰[1] 王俊文[1] 戴跃伟[1] 王执铨[1] 

机构地区：[1]南京理工大学自动化学院,江苏南京210094 [2]重庆电子工程职业学院计算机应用系,重庆401331

出　　处：《电子学报》2010年第11期2578-2584,共7页Acta Electronica Sinica

基　　金：江苏省自然科学基金(No.BK201022321);江苏省自然科学基金(No.BK2008403);中国博士后基金(No.20070421017);教育部博士点基金(No.20093219120037);南京理工大学自主科研计划(No.2010ZYTS048)

摘　　要：扩频隐写分析是信息隐藏研究领域的一个重要方面.文章提出基于高阶Markov链的数字图像统计分布模型,在对常用图像扫描方法构成高阶Markov链的效果进行比较后,采用Hilbert扫描方式构建数字图像n阶Markov链模型,进而提出度量数字图像隐写统计安全性的n阶Markov链测度,并证明其有界.最后文章通过研究扩频隐写对高阶Markov链模型经验矩阵的影响,利用该模型提取图像统计特征,并使用支持向量机对几种常用图像扩频隐写方法进行分析.实验说明文章所提方法对扩频隐写分析效果良好,且随着模型阶数提高,分析准确率也随之提高.Steganalysis of spread spectrum image steganography is an important aspect in the domain of information hiding research.An image statistical distribution model based on high-order Markov chain was proposed.By a comparison on the effect of n-order Markov chain achieved by several general image scanning methods,Hilbert scanning method was used to construct the proposed model.After that,a n-order statistic security measure for digital image steganography based on the model was proposed,and that it is bounded was proved.Finally,after the effect of spread spectrum steganography on empirical matrix of the proposed model was studied,image statistical features which based on the proposed model were extracted,and several common spread spectrum steganography algorithms were analyzed using support vector machine.Experiment shows the active effects of steganalysis of spread spectrum image steganography based on the proposed model is obvious,furthermore,the steganalysis accuracy is improved as the model order increased.

关 键 词：信息隐藏 隐写分析 高阶马尔科夫链 扩频图像隐写 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]