图像隐写定量与定位分析关键问题研究
|
摘要
隐写的定量和定位分析是隐写分析技术研究的热点和难点。目前，隐写的定量和定位分析方面已有许多优秀的成果，但距离实际应用需求还存在很大差距。本文主要针对数字图像隐写的定量和定位分析中的若干关键问题展开研究，全文共七章，主要包括三个方面：
一、背景知识及国内外研究进展方面。从应用背景和技术背景阐述了隐写的定量与定位分析的现实意义和理论价值；简要介绍了数字隐写和隐写分析的研究概况；详细阐述了图像隐写的定量与定位分析方面的研究进展，并分析指出了当前存在的一些主要问题。
二、图像隐写的定量分析研究方面。针对MLSB(Multiple Least Significant Bits)替换隐写的低嵌入率估计问题，基于像素组间的转移关系，给出了TMLSB替换隐写(TypicalMLSB replacement)和IMLSB替换隐写(Independent MLSB replacement)的定量分析方法；针对MLSB替换隐写的高嵌入率估计问题，基于带权隐密图像，给出了IMLSB替换隐写和ID-MLSB替换隐写(Independent MLSB replacement with possible Differnent ratios)的定量分析方法；针对图像特征间差异的刻画问题，基于相对熵，给出了直方图类特征差异计算方法，且以该计算方法为基础，给出了基于相对熵度量直方图类特征差异的JPEG图像隐写的定量分析方法。其中：
1.基于像素组跟踪的MLSB替换隐写的定量分析。首先，将图像用相邻像素组描述，基于MLSB替换隐写的像素组跟踪集和跟踪子集，给出了像素组间的转移关系；其次，基于两类对称掩码作用下像素组噪声水平的变化，指出了原始图像和隐密图像的一些统计特性；然后，基于上述像素组转移关系和图像统计特性，给出了TMLSB替换隐写和IMLSB替换隐写的定量分析方法。实验结果表明：所给出方法对低嵌入率的估计精度明显高于已有的定量分析方法。
2.基于带权隐密图像的MLSB替换隐写的定量分析。首先，基于IMLSB替换隐写的带权隐密图像，给出了IMLSB替换隐写的定量分析方法，指出了基于带权隐密图像的IMLSB替换隐写和TMLSB替换隐写的定量分析方法间的关系；然后，基于ID-MLSB替换隐写的带权隐密图像，给出了ID-MLSB替换隐写的定量分析方法。实验结果表明：在嵌入率较大，尤其是接近1时，基于带权隐密图像的定量分析方法对嵌入率的估计误差明显小于现有典型的定量分析方法。
3.基于相对熵度量直方图类特征差异的JPEG隐写定量分析。首先，基于两假设下的最优检验——似然比检验，给出了一种基于相对熵的直方图类特征差异计算方法；然后，基于该计算方法和机器学习的思想，给出了基于相对熵度量直方图类特征差异的JPEG隐写定量分析方法。实验结果表明：针对JSteg、F5、以及-F5和nsF5等改进的F5隐写，与典型的基于差值度量直方图类特征差异的JPEG隐写定量分析方法相比，基于相对熵度量直方图类特征差异的JPEG隐写定量分析方法对系数更改比率的估计误差要小得多，而且受系数更改比率的影响相对较小。
三、图像隐写的定位分析研究方面。针对多幅嵌入路径相同的隐密图像，基于样本块选取和定量分析，给出了定位分析的一般方法；针对单幅TMLSB替换隐写的隐密图像，基于带权隐密像素，分析了TMLSB替换隐写的性质，给出了基于最小和子序列的顺序TMLSB替换隐写的定位分析方法；最后，结合碰撞攻击的思想，给出了基于定位分析的隐写密钥还原方法。其中：
1.基于样本块选取和定量分析的定位分析。针对多幅嵌入路径相同的隐密图像，首先，将待判定位置是否隐密的判定问题转化为待判定位置的嵌入率估计问题；其次，基于样本块选取，给出了待判定位置中的嵌入率估计方法；然后，根据待判定位置中的定量分析结果，给出了定位分析的一般方法；最后，以该一般方法为基础，给出了多个LSB替换隐写和TMLSB替换隐写的定位分析算法。理论分析和实验结果表明：给出的一般方法不仅将Ker的LSB替换隐写的定位分析算法纳入其中，而且能够用于TMLSB替换隐写的定位分析；此外，随着嵌入路径相同的隐密图像的增多，根据该一般方法设计的定位分析算法对嵌入位置的定位准确率将得到大幅度提高。
2.基于最小和子序列的TMLSB替换隐写的定位分析。针对单幅TMLSB替换隐写的隐密图像，首先，分析并指出了将带权隐密图像的思想应用于TMLSB替换隐写的定位分析须要解决的问题；在此基础上，分析了基于带权隐密像素的TMLSB替换隐写的性质；然后，基于该性质，将顺序TMLSB替换隐写的起止位置估计转化为最小和子序列的寻找问题，给出了基于最小和子序列的顺序TMLSB替换隐写的定位分析方法。实验结果表明，所给出方法能够以较高的精度估计出顺序TMLSB替换隐写的隐密像素起止位置。而且，该方法还将Ker和B hme的顺序LSB替换隐写的隐密像素起止位置估计算法纳入其中，适用范围更广。
3.基于定位分析的隐写密钥还原。首先，基于碰撞攻击的思想，给出了基于定位分析的隐写密钥还原方法；其次，基于二项分布的正态近似，分析了取伪率、弃真率、嵌入率、定位分析的准确性等若干因素对隐写密钥还原的影响；然后，结合基于改进WS(Weighted Stego Image)的LSB替换隐写的定位分析算法和基于l-WS的TMLSB替换隐写的定位分析算法，分别给出基于改进WS的LSB替换隐写的密钥还原算法和基于l-WS的TMLSB替换隐写的密钥还原算法。实验结果表明，本文方法不仅在隐写密钥还原的正确率和速度方面得到了明显改善，而且能够结合定位分析方面的研究结果对更多隐写的密钥进行还原。
最后，对全文工作进行了总结，并对下一步的研究方向进行了展望。
Quantitative and locating steganalysis are the hot and difficult issues in the field ofsteganalysis. Currently, there have been numerous excellent results on quantitative and locatingsteganalysis, but which still can not satisfy the requirements of steganalyzers. This thesis mainlydiscusses the researches on quantitative and locating image steganalysis, includes seven chapterswhich can be summarized into the following three parts:
The first part: background knowledge and state of the art. As viewed from the applicationand technical background, the practical and theoretical values of steganalysis are described. Theresearch review of digital steganography and steganalysis are introduced briefly. The state of theart in quantitative and locating image steganalysis are surveyed in detail. And some key issues onquantitative and locating image steganalysis are pointed out.
The second part: the quantitative image steganalysis. In regard to the problem of estimatingthe low embedding ratio of MLSB (Multiple Least Significant Bits) replacement steganography,two quantitative steganalysis methods are respectively proposed for TMLSB replacement(Typical MLSB replacement) steganography and IMLSB replacement (Independent MLSBreplacement) steganography based on the transition relationship among pixel groups; in regard tothe problem of estimating the high embedding ratio of MLSB replacement steganography, twoquantitative steganalysis methods are respectively proposed for IMLSB replacement and ID-MLSB replacement (Independent MLSB replacement with Different ratios) steganography basedon the weighted stego image; in response to the problem of measuring the difference betweenimage features, based on relative entropy, a method is proposed to compute the differencebetween histogram-like features, then based on which, a quantitative steganalysis method isproposed for JPEG image steganography.
1. Quantitative steganalysis of MLSB replacement steganography based on pixel grouptrace. First, the adjacent pixel groups are used to described an image and the transitionrelationship among pixel groups are given based on the pixel group trace set and trace subset ofMLSB replacement; second, based on the change of the pixel group’s noise level when applyingtwo categories of symmetrical masks, some statistical characteristics of the cover and stegoimages are pointed out; third, based on above transition relationship and statistical characteristics,two quantitative steganalysis methods are proposed for TMLSB replacement and IMLSBreplacement steganography respectively. Experimental results show that the proposedquantitative steganalysis methods significantly outperform other methods for low embeddingratio.
2. Quantitative steganalysis of MLSB replacement steganography based on weighted stegoimage. First, based on the weighted stego image of IMLSB replacement, a quantitativesteganalysis method is proposed for IMLSB replacement, and the relationship between theproposed method and the weighted stego image steganalysis method for TMLSB replacement ispointed out; then, based on the weighted stego image of ID-MLSB replacement, a quantitativesteganalysis method is proposed for ID-MLSB replacement steganography. Experimental results show that when the embedding ratio is high, especially when the embedding ratio is close to1,the proposed quantitative steganalysis methods own significantly smaller error than others.
3. Quantitative steganalysis of JPEG image steganography via measuring the differencebetween histogram-like features based on relative entropy. First, based on the optimum twohypotheses test---likelihood ratio test, a relative-entropy-based method is proposed to computethe difference between histogram-like features; then, based on the proposed differencecomputing method and the idea of machine learning, a quantitative steganalysis method isproposed for JPEG image steganography via measuring the difference between histogram-likefeatures based on relative entropy. Experimental results show that for JSteg, F5and somevariants of F5, such as-F5and nsF5, compared with the typical quantitative JPEG steganalysismethod which directly computing the diffence between two histogram-like features, the proposedquantitative JPEG image steganalysis can estimate the modification ratio with smaller error, andwould be affected by the actual modification ratio less.
The third part: the locating image steganalysis. For the case of owning multiple stegoimages which are embedded message into along the same embedding path, a general locatingsteganalysis method is proposed based on sample block selection and quantitative steganalysis;for the case of owning single stego image of TMLSB replacement, a property of TMLSBreplacement is analyzed based on the weighted stego pixel, and a locating steganalysis methodfor sequential TMLSB replacement steganography is proposed based on minimum sumsubsequence; for the problem of stego key recovery, from the idea of collision attack, a reversingmethod is proposed based on locating steganalysis.
1. Locating steganalysis based on sample block selection and quantitative steganalysis. Forthe case of owning multiple stego image with the same embedding path, first, the problem ofdetermine the stego positions is converted into estimating the embedding ratios in the positionsfor determining; second, based on sample block selection, a method is proposed to estimate theembedding ratio in each position; third, based on the estimated embedding ratio, a generallocating steganalysis is proposed; finally, based on the general method, some locatingsteganalysis algorithms are proposed for LSB replacement steganography and TMLSBreplacement steganography. Theoretic analysis and experimental results show that the proposedalgorithm not only contains Ker’s locating steganalysis algorithm for LSB replacement, but alsocan be to locate the stego pixels of TMLSB replacement; and with the increase of the number ofstego images with the same embedding path, the exactness of the proposed locating steganalysisalgorithms will be improved significantly.
2. Locating steganalysis of TMLSB replacement based on minimum sum subsequence. Forthe case of owning single stego image of TMLSB replacement, first, the problems necessary tobe settled when applying the idea of weighted stego image to the locating steganalysis ofTMLSB replacement are pointed out; second, a property of TMLSB replacement is analyzedbased on weighted stego pixel; third, based on the property, the process of estimating the startand end positions of sequential TMLSB replacement is converted into the problem of findingminimum sum subsequence, and the locating steganalysis algorithms for sequential TMLSB replacement are propsoed. Experimental results show that the proposed locating steganalysismethod for sequential TMLSB replacement can estimate the start and end positions with highaccuracy. Additinally, the proposed method contains the Ker and B hme’s algorithms forsequential LSB replacement, owns wider applicability.
3. Stego key recovery based on locating steganalysis. First, from the idea of collision attack,a stego key recovery method is proposed based on locating steganalysis; second, based on thenormal approximation of binomial distribution, the effects of false alarm rate, miss alarm rate,embedding ratio and accuracy of locating steganalysis on the efficiency of stego key reversing isanalyzed; third, a stego key recovery algorithm for LSB replacement is given based on theimproved weighted stego image method, and a steg key recovery algorithm for TMLSBreplacement is given based on the weighted stego image method. Experimental results show thatthe proposed method can recover the correct stego key with higher ratio and improve the speedof recovery. Additionally, the proposed stego key recovery method can utilize the results oflocating steganalysis to recover the stego key for more steganography.
Finally, a conclusion with a discussion of the direction for the future research is given.
一、背景知识及国内外研究进展方面。从应用背景和技术背景阐述了隐写的定量与定位分析的现实意义和理论价值；简要介绍了数字隐写和隐写分析的研究概况；详细阐述了图像隐写的定量与定位分析方面的研究进展，并分析指出了当前存在的一些主要问题。
二、图像隐写的定量分析研究方面。针对MLSB(Multiple Least Significant Bits)替换隐写的低嵌入率估计问题，基于像素组间的转移关系，给出了TMLSB替换隐写(TypicalMLSB replacement)和IMLSB替换隐写(Independent MLSB replacement)的定量分析方法；针对MLSB替换隐写的高嵌入率估计问题，基于带权隐密图像，给出了IMLSB替换隐写和ID-MLSB替换隐写(Independent MLSB replacement with possible Differnent ratios)的定量分析方法；针对图像特征间差异的刻画问题，基于相对熵，给出了直方图类特征差异计算方法，且以该计算方法为基础，给出了基于相对熵度量直方图类特征差异的JPEG图像隐写的定量分析方法。其中：
1.基于像素组跟踪的MLSB替换隐写的定量分析。首先，将图像用相邻像素组描述，基于MLSB替换隐写的像素组跟踪集和跟踪子集，给出了像素组间的转移关系；其次，基于两类对称掩码作用下像素组噪声水平的变化，指出了原始图像和隐密图像的一些统计特性；然后，基于上述像素组转移关系和图像统计特性，给出了TMLSB替换隐写和IMLSB替换隐写的定量分析方法。实验结果表明：所给出方法对低嵌入率的估计精度明显高于已有的定量分析方法。
2.基于带权隐密图像的MLSB替换隐写的定量分析。首先，基于IMLSB替换隐写的带权隐密图像，给出了IMLSB替换隐写的定量分析方法，指出了基于带权隐密图像的IMLSB替换隐写和TMLSB替换隐写的定量分析方法间的关系；然后，基于ID-MLSB替换隐写的带权隐密图像，给出了ID-MLSB替换隐写的定量分析方法。实验结果表明：在嵌入率较大，尤其是接近1时，基于带权隐密图像的定量分析方法对嵌入率的估计误差明显小于现有典型的定量分析方法。
3.基于相对熵度量直方图类特征差异的JPEG隐写定量分析。首先，基于两假设下的最优检验——似然比检验，给出了一种基于相对熵的直方图类特征差异计算方法；然后，基于该计算方法和机器学习的思想，给出了基于相对熵度量直方图类特征差异的JPEG隐写定量分析方法。实验结果表明：针对JSteg、F5、以及-F5和nsF5等改进的F5隐写，与典型的基于差值度量直方图类特征差异的JPEG隐写定量分析方法相比，基于相对熵度量直方图类特征差异的JPEG隐写定量分析方法对系数更改比率的估计误差要小得多，而且受系数更改比率的影响相对较小。
三、图像隐写的定位分析研究方面。针对多幅嵌入路径相同的隐密图像，基于样本块选取和定量分析，给出了定位分析的一般方法；针对单幅TMLSB替换隐写的隐密图像，基于带权隐密像素，分析了TMLSB替换隐写的性质，给出了基于最小和子序列的顺序TMLSB替换隐写的定位分析方法；最后，结合碰撞攻击的思想，给出了基于定位分析的隐写密钥还原方法。其中：
1.基于样本块选取和定量分析的定位分析。针对多幅嵌入路径相同的隐密图像，首先，将待判定位置是否隐密的判定问题转化为待判定位置的嵌入率估计问题；其次，基于样本块选取，给出了待判定位置中的嵌入率估计方法；然后，根据待判定位置中的定量分析结果，给出了定位分析的一般方法；最后，以该一般方法为基础，给出了多个LSB替换隐写和TMLSB替换隐写的定位分析算法。理论分析和实验结果表明：给出的一般方法不仅将Ker的LSB替换隐写的定位分析算法纳入其中，而且能够用于TMLSB替换隐写的定位分析；此外，随着嵌入路径相同的隐密图像的增多，根据该一般方法设计的定位分析算法对嵌入位置的定位准确率将得到大幅度提高。
2.基于最小和子序列的TMLSB替换隐写的定位分析。针对单幅TMLSB替换隐写的隐密图像，首先，分析并指出了将带权隐密图像的思想应用于TMLSB替换隐写的定位分析须要解决的问题；在此基础上，分析了基于带权隐密像素的TMLSB替换隐写的性质；然后，基于该性质，将顺序TMLSB替换隐写的起止位置估计转化为最小和子序列的寻找问题，给出了基于最小和子序列的顺序TMLSB替换隐写的定位分析方法。实验结果表明，所给出方法能够以较高的精度估计出顺序TMLSB替换隐写的隐密像素起止位置。而且，该方法还将Ker和B hme的顺序LSB替换隐写的隐密像素起止位置估计算法纳入其中，适用范围更广。
3.基于定位分析的隐写密钥还原。首先，基于碰撞攻击的思想，给出了基于定位分析的隐写密钥还原方法；其次，基于二项分布的正态近似，分析了取伪率、弃真率、嵌入率、定位分析的准确性等若干因素对隐写密钥还原的影响；然后，结合基于改进WS(Weighted Stego Image)的LSB替换隐写的定位分析算法和基于l-WS的TMLSB替换隐写的定位分析算法，分别给出基于改进WS的LSB替换隐写的密钥还原算法和基于l-WS的TMLSB替换隐写的密钥还原算法。实验结果表明，本文方法不仅在隐写密钥还原的正确率和速度方面得到了明显改善，而且能够结合定位分析方面的研究结果对更多隐写的密钥进行还原。
最后，对全文工作进行了总结，并对下一步的研究方向进行了展望。
Quantitative and locating steganalysis are the hot and difficult issues in the field ofsteganalysis. Currently, there have been numerous excellent results on quantitative and locatingsteganalysis, but which still can not satisfy the requirements of steganalyzers. This thesis mainlydiscusses the researches on quantitative and locating image steganalysis, includes seven chapterswhich can be summarized into the following three parts:
The first part: background knowledge and state of the art. As viewed from the applicationand technical background, the practical and theoretical values of steganalysis are described. Theresearch review of digital steganography and steganalysis are introduced briefly. The state of theart in quantitative and locating image steganalysis are surveyed in detail. And some key issues onquantitative and locating image steganalysis are pointed out.
The second part: the quantitative image steganalysis. In regard to the problem of estimatingthe low embedding ratio of MLSB (Multiple Least Significant Bits) replacement steganography,two quantitative steganalysis methods are respectively proposed for TMLSB replacement(Typical MLSB replacement) steganography and IMLSB replacement (Independent MLSBreplacement) steganography based on the transition relationship among pixel groups; in regard tothe problem of estimating the high embedding ratio of MLSB replacement steganography, twoquantitative steganalysis methods are respectively proposed for IMLSB replacement and ID-MLSB replacement (Independent MLSB replacement with Different ratios) steganography basedon the weighted stego image; in response to the problem of measuring the difference betweenimage features, based on relative entropy, a method is proposed to compute the differencebetween histogram-like features, then based on which, a quantitative steganalysis method isproposed for JPEG image steganography.
1. Quantitative steganalysis of MLSB replacement steganography based on pixel grouptrace. First, the adjacent pixel groups are used to described an image and the transitionrelationship among pixel groups are given based on the pixel group trace set and trace subset ofMLSB replacement; second, based on the change of the pixel group’s noise level when applyingtwo categories of symmetrical masks, some statistical characteristics of the cover and stegoimages are pointed out; third, based on above transition relationship and statistical characteristics,two quantitative steganalysis methods are proposed for TMLSB replacement and IMLSBreplacement steganography respectively. Experimental results show that the proposedquantitative steganalysis methods significantly outperform other methods for low embeddingratio.
2. Quantitative steganalysis of MLSB replacement steganography based on weighted stegoimage. First, based on the weighted stego image of IMLSB replacement, a quantitativesteganalysis method is proposed for IMLSB replacement, and the relationship between theproposed method and the weighted stego image steganalysis method for TMLSB replacement ispointed out; then, based on the weighted stego image of ID-MLSB replacement, a quantitativesteganalysis method is proposed for ID-MLSB replacement steganography. Experimental results show that when the embedding ratio is high, especially when the embedding ratio is close to1,the proposed quantitative steganalysis methods own significantly smaller error than others.
3. Quantitative steganalysis of JPEG image steganography via measuring the differencebetween histogram-like features based on relative entropy. First, based on the optimum twohypotheses test---likelihood ratio test, a relative-entropy-based method is proposed to computethe difference between histogram-like features; then, based on the proposed differencecomputing method and the idea of machine learning, a quantitative steganalysis method isproposed for JPEG image steganography via measuring the difference between histogram-likefeatures based on relative entropy. Experimental results show that for JSteg, F5and somevariants of F5, such as-F5and nsF5, compared with the typical quantitative JPEG steganalysismethod which directly computing the diffence between two histogram-like features, the proposedquantitative JPEG image steganalysis can estimate the modification ratio with smaller error, andwould be affected by the actual modification ratio less.
The third part: the locating image steganalysis. For the case of owning multiple stegoimages which are embedded message into along the same embedding path, a general locatingsteganalysis method is proposed based on sample block selection and quantitative steganalysis;for the case of owning single stego image of TMLSB replacement, a property of TMLSBreplacement is analyzed based on the weighted stego pixel, and a locating steganalysis methodfor sequential TMLSB replacement steganography is proposed based on minimum sumsubsequence; for the problem of stego key recovery, from the idea of collision attack, a reversingmethod is proposed based on locating steganalysis.
1. Locating steganalysis based on sample block selection and quantitative steganalysis. Forthe case of owning multiple stego image with the same embedding path, first, the problem ofdetermine the stego positions is converted into estimating the embedding ratios in the positionsfor determining; second, based on sample block selection, a method is proposed to estimate theembedding ratio in each position; third, based on the estimated embedding ratio, a generallocating steganalysis is proposed; finally, based on the general method, some locatingsteganalysis algorithms are proposed for LSB replacement steganography and TMLSBreplacement steganography. Theoretic analysis and experimental results show that the proposedalgorithm not only contains Ker’s locating steganalysis algorithm for LSB replacement, but alsocan be to locate the stego pixels of TMLSB replacement; and with the increase of the number ofstego images with the same embedding path, the exactness of the proposed locating steganalysisalgorithms will be improved significantly.
2. Locating steganalysis of TMLSB replacement based on minimum sum subsequence. Forthe case of owning single stego image of TMLSB replacement, first, the problems necessary tobe settled when applying the idea of weighted stego image to the locating steganalysis ofTMLSB replacement are pointed out; second, a property of TMLSB replacement is analyzedbased on weighted stego pixel; third, based on the property, the process of estimating the startand end positions of sequential TMLSB replacement is converted into the problem of findingminimum sum subsequence, and the locating steganalysis algorithms for sequential TMLSB replacement are propsoed. Experimental results show that the proposed locating steganalysismethod for sequential TMLSB replacement can estimate the start and end positions with highaccuracy. Additinally, the proposed method contains the Ker and B hme’s algorithms forsequential LSB replacement, owns wider applicability.
3. Stego key recovery based on locating steganalysis. First, from the idea of collision attack,a stego key recovery method is proposed based on locating steganalysis; second, based on thenormal approximation of binomial distribution, the effects of false alarm rate, miss alarm rate,embedding ratio and accuracy of locating steganalysis on the efficiency of stego key reversing isanalyzed; third, a stego key recovery algorithm for LSB replacement is given based on theimproved weighted stego image method, and a steg key recovery algorithm for TMLSBreplacement is given based on the weighted stego image method. Experimental results show thatthe proposed method can recover the correct stego key with higher ratio and improve the speedof recovery. Additionally, the proposed stego key recovery method can utilize the results oflocating steganalysis to recover the stego key for more steganography.
Finally, a conclusion with a discussion of the direction for the future research is given.
引文
[1]沈昌祥,张焕国,冯登国,曹珍富,黄继武.信息安全综述[J].中国科学E辑:信息科学,2007,37(2):129-150.
[2] Herodotus. The Histories [M]. London, England: J. M. Dent&Sons, Ltd,1992.
[3] Kurak C, McHugh J. A cautionary note on image downgrading [A]. In: Proceedings of the8th IEEE Annual Computer Security Applications Conference [C],1992,153-159.
[4] Simmons G J. The prisoners’ problem and the subliminal channel [A]. In: Proceedings ofCRYPTO [C],1983,51-67.
[5] Cachin C. An information-theoretic model for steganography [A]. In: Proceedings of2thInternational Workshop on Information Hiding [C],1998, LNCS, vol.1525,306-318.
[6] Cachin C. An information-theoretic model for steganography [J]. Information andComputation,2004,192(1):41-56.
[7] Hopper N J, Langford J, Ahn L V. Provably secure steganography [A]. In: Proceedings pfCRYPTO [C],2002, LNCS, vol.2442,77-92.
[8]刘光杰,戴跃伟,赵玉鑫,王执铨.隐写对抗的博弈论建模[J].南京理工大学(自然科学版),2008,32(2):199-204.
[9] Fridrich J, Goljan M. Digital image steganography using stochastic modulation [A]. In:Proceedings of SPIE, Security and Watermarking of Multimedia Contents V [C].2003,SPIE, vol.5020,191-202.
[10]Yang C, Liu F, S Lian, Luo X. Weighted stego-image steganalysis of message hidden intoeach bit plane [J]. The Computer Journal,2011, doi:10.1093/comjnl/bxr112.
[11]Fridrich J, Soukal D, Goljan M. Maximum likelihood estimation of length of secret messageembedded using+-K steganography in spatial domain [A]. In: Proceedings of SPIE,Electronic Imaging, Security, Steganography and Watermarking of Multimedia Contents VII
[C],2005, SPIE vol.5681,595-606.
[12]Fridrich J, Rui D. Secure steganographic methods for palette images [A]. In: Proceedings of3rd International Workshop on Information Hiding [C], LNCS, vol.1768,2000,47-60.
[13]Pan H K, Chen Y Y, Tseng Y C. A secure data hiding scheme for two color images [A]. In:Proceedings of IEEE Symposium on Computer and Communications [C],2000,750-755.
[14]Tseng Y C, Pan H K. Data hiding in2-color images [J]. IEEE Transactions on Computers.2002,51(7):873-878.
[15]Kawaguchi E, Eason R O. Principle and applications of BPCS-steganography [A]. In:Proceedings of SPIE, Multimedia Systems and Applications [C],1998, SPIE vol.3528,464-473.
[16]Nguyen B, Yoon S, Lee H. Multi bit plane image steganography [A]. In: Proceeding ofInternational Workshop on Digital Watermarking [C],2006, LNCS, vol.4283,61-70.
[17]Luo W, Huang F, Huang J. Edge adative image steganography based on LSB matchingrevisited [J]. IEEE Transactions on Information Forensics and Security,5(2):201-214.
[18]Katzenbeisser S, Petitcolas F A P编.吴秋新,钮心忻,杨义先等译.信息隐藏技术—隐写术与数字水印[M].北京:人民邮电出版社,2001.
[19]Fridrich J, Pevny. Statistically undetectable JPEG steganography: dead ends challenges, andopportunities [A]. In: Proceedings of9th Workshop on Multimedia&Security [C],2007,3-14.
[20]Provos N. Defending against statistical steganalysis [A]. In: Proceedings of10th USENIXSecurity Symposium [C],2001,323-335.
[21]Hetzl S, Mutzel. A graph-theoretic approach to steganography [A]. In: Proceedings of9thIFIP TC-6TC-11International Conference on Communications and Multimedia Security,Lecture Notes in Computer Science [C],2005, vol.3677,119-128.
[22]Sallee P. Model-based steganography [A]. In: Proceedings of International Workshop onDigital Watermarking [C],2003, LNCS, vol.2939,154-167.
[23]Sallee P. Model-based methods for steganography and steganalysis [J]. International Journalof Image Graphics,2005,5(1):167-190.
[24]Westfeld A. High capacity despite better steganalysis (F5-a steganographic algorithm)[A].In: Proceedings of4th International Workshop on Information Hiding [C],2001, LNCS, vol.2137,289-302.
[25]Kim Y, Duric Z, Richards D. Modified matrix encoding technique for minimal distortionsteganography. Information Hiding [A]. In: Proceedings of8th International Workshop onInformation Hiding [C],2006, LNCS, vol.4437,314-327.
[26]Fridrich J, Goljan M, Soukal D. Perturbed quantization steganography [J]. ACM MultimediaSystem Journal,2005,11(2):98-107.
[27]Low S H, Maxemchuk N F, Lapone A M. Document identification for copyright protectionusing centroid detection [J]. IEEE Transaction on Communications,1998,46(3):372-383.
[28]刘东,孙明,周明天.基于图论的文本数字水印技术[J].计算机研究与发展,2007,44(10):1757-1764.
[29]Wayner P. Mimic Functions [J]. Cryptologia,1992, XVI(3):193-214.
[30]Wayner P. Strong theoretical steganography [J]. Cryptologia,1995, XIX(3):285-299.
[31]Fridrich J, Goljan M. Practical steganalysis of digital images—state of the art [A]. In:Proceedings of SPIE, Security and Watermarking of Multimedia Contents IV [C],2002,SPIE, vol.4675,1-13.
[32]钮心忻.信息隐藏与数字水印[M].北京:北京邮电大学出版社,2004.
[33]Luo X, Wang D, Wang P, Liu F. A review on blind detection for image steganography [J].Signal Processing,2008,88(9):2138-2157.
[34]Avcibas, Memon N, Sankur B. Steganalysis using image quality metrics [J]. IEEETransactions on Image Processing,2003,12(2):221-229.
[35]Farid H. Detecting steganographic messages in digital images [R]. Hanover, NH: DartmouthCollege,2001.
[36]Goljan M, Fridrich J, Holotyak T. New blind steganalysis and its implications [A]. In:Proceedings of SPIE, Security, Steganography, and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol.6072,1-13.
[37]Harmsen J J, Pearlman W A. Steganalysis of additive noise modelable information hiding
[A]. In: Proceedings of SPIE, Security, Steganography, and Watermarking of MultimediaContents V [C],2003, SPIE, vol.5020,131-142.
[38]Shi Y, Xuan G, Yang C, Gao J, and et al. Effective steganalysis based on statistical momentsof wavelet characteristic function [A]. In: Proceedings of IEEE International Conference onInformation Technology: Coding and Computing [C],2005,768-773.
[39]Chen X C, Wang Y H, Tan T N, Guo L. Blind image steganalysis based on statisticalanalysis of empirical matrix [A]. In: Proceedings of18th International Conference onPattern Recognition [C],2006, vol.3,1107-1110.
[40]Lie W N, Lin G S. A feature-based classification technique for blind image steganalysis [J].IEEE Transactions on Multimedia,2005,7(6):1007-1020.
[41]Pevny T, Fridrich J. Towards multi-class blind steganalyzer for JPEG images [A]. In:Proceedings of4th International Workshop on Digital Watermarking [C],2005, LNCS, vol.3710,39-53.
[42]Pevny T, Fridrieh J. Multiclass detector of current steganographic methods for JPEG format[J]. IEEE Transactions on Information Forensics and Security,2008,3(4):635-650.
[43]Dong J, Wang W, Tan T N. Multi-class blind steganalysis based on image run-lengthanalysis. In: Proceedings of7th International Workshop on Digital Watermarking [C],2009,LNCS, vol.5703,199-210.
[44]Chiew K L, Josef P. Binary Image steganographic techniques classification based on multi-class steganalysis [A]. In: Proceedings of6th International Conference on InformationSecurity, Practice and Experience [C],2010, LNCS, vol.6047,341-358.
[45]Fridrich J, Goljan M, Hogea D, Souka D. Quantitative steganalysis of digital images:estimating the secret message length [J]. ACM Multimedia Systems Journal, Special issueon Multimedia Security,2003,9(3):288-302.
[46]Dabeer O, Sullian K, Madhow U, et al. Detection of hiding in the least significant bit [J].IEEE Transactions on Signal Processing,2004,52(10):3046-3058.
[47]Fridrich J, Goljan M, Hogea D. Steganalysis of JPEG images: breaking the F5algorithm [A].In: Proceedings of5th International Workshop on Information Hiding [C],2002, LNCS, vol.258,310-323.
[48]Fridrich J, Goljan M, Du R. Reliable detection of LSB steganography in color and grayscaleimages [A]. In: Proceedings of Workshop on Multimedia&Security [C],2001,27-30.
[49]Dumitrescu S, Wu X, Wang Z. Detection of LSB steganography via sample pair analysis [J].IEEE Transactions on Signal Processing,2003,51(7):1995-2007.
[50]Fridrich J, Goljan M. On estimation of secret message length in LSB steganography inspatial domain [A]. In: Proceedings of SPIE Security, Steganography and Watermarking ofMultimedia Contents VI [C],2004, SPIE, vol.5306,23-24.
[51]Ker A D, B hme R. Revisiting weighted stego-image steganalysis [A]. In: Proceedings ofSPIE Security, Steganography and Watermarking of Multimedia Contents X [C],2007,SPIE, vol.6819,27-31.
[52]Fridrich J, Soukal D, Goljan M. Maximum likelihood estimation of length of secret messageembedded using+-K steganography in spatial domain [A]. In: Proceedings of SPIE,Electronic Imaging, Security, Steganography and Watermarking of Multimedia Contents VII
[C],2005, SPIE, vol.5681,595-606.
[53]He J, Huang J. Steganalysis of stochastic modulation steganography [J]. Science in China:Series F Information Sciences,2006,49(3):273-285.
[54]Pevny T, Fridrich J, Ker A D. From blind to quantitative steganalysis [A]. In: Proceedings ofSPIE, Electronic Imaging, Security and Forensics of Multimedia Contents XI [C],2009,SPIE, vol.7254,0C-1-0C-14.
[55]Guo Y, Kong X, You X, Li L. Universal methodology for developing quantitativesteganalysis [J]. Chinese Journal of Electronics,2009,18(3):455-459.
[56]Trivedi S, Chandramouli R. Secret key estimation in sequential steganography [J]. IEEETransactions on Signal Processing,2005,53(2):746-757.
[57]Kong X, Liu W, You X. Secret message location steganalysis based on local coherences ofhue [A]. In: Proceeding of Pacific Rim Conference on Multimedia Part II [C],2005, LNCS,vol.3768,301-311.
[58]Ambalavanan A, Chandramouli R. A Bayesian image steganalysis approach to estimate theembedded secret message [A]. In: Proceedings of7th ACM Workshop on Multimedia andSecurity [C],2005,33-38.
[59]Davidson I, Paul G. Locating secret messages in images [A]. In: Proceedings of10th ACMSIGKDD International Conference on Knowledge Discovery and Data Mining [C],2004,545-550.
[60]Ker A D. Locating steganographic payload via WS residuals [A]. In: Proceedings of10thWorkshop on Multimedia and Security [C],2008,27-31.
[61]Chiew K L, Pieprzyk J. Identifying steganographic payload location in binary image [A]. In:Proceeding of Pacific Rim Conference on Multimedia Part I [C],2010, LNCS, vol.6297,590-600.
[62]Provos N, Honeyman P. Detecting steganographic content on the Internet [R]. InternetSociety, Network and Distributed System Security Symposium,2002.
[63]张卫明,李世取,刘九芬.隐写术CPT的等价密钥分析[J].电子学报,2007,35(12):2258-2261.
[64]Fridrich J, Goljan M, Soukal D. Searching for the stego-key [A]. In: Proceedings of SPIESecurity, Steganography, and Watermarking of Multimedia Contents VII [C],2004, SPIE,vol.5306,70-82.
[65]Fridrich J, Goljan M, Soukal D, Holotyak T. Forensic steganalysis: determining the stegokey in spatial domain steganography [A]. In: Proceedings of SPIE Security, Steganography,and Watermarking of Multimedia Contents VII [C],2005, SPIE vol.5681,631-642.
[66]张卫明,李世取,刘九芬.对空域图像LSB隐写术的提取攻击[J].计算机学报,2007，30(9):1625-1631.
[67]张卫明.隐写信息提取的理论与方法研究[D].信息工程大学,2005.
[68]张卫明,刘九芬,李世取. LSB隐写术的密钥恢复方法[J].中山大学学报(自然科学版),2005,44(3):29-33.
[69]Zhang T, Ping X J. A fast and effective steganalytic technique against Jsteg-like algorithms
[A]. In: Proceedings of Symposium on Applied Computing [C],2003,307-311.
[70]Zhang T, Ping X J. A new approach to reliable detection of LSB steganography in naturalimage [J]. Signal Process,2003,83(10):545-548.
[71]Lu P, Luo X, Tang Q, Zou Y. An improved sample pairs method for detection of LSBembedding [A]. In: Proceedings of6th Workshop on Information Hiding [C],2004, LNCS,vol.3200,116-127.
[72]Ker A D. A general framework for the structural steganalysis of LSB replacement [A]. In:Proceedings of7th International Workshop on Information Hiding [C],2005, LNCS, vol.3727:296-311.
[73]Ker A D. Fourth-order structural steganalysis and analysis of cover assumptions [A]. In:Proceedings of SPIE, Security, Steganography and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol.6072:25-38.
[74]Dumitrescu S, Wu X. A new framework of LSB steganalysis of digital media [J]. IEEETransactions on Signal Process,2005,53(10):3936-3947.
[75]Luo X, Yang C, Liu F. Equivalence analysis among DIH, SPA, and RS steganalysismethods [A]. In: Proceedings of the10th IFIP conference of Communication andMultimedia Security [C],2006, LNCS, vol.4237,161-172.
[76]张涛.图象隐写分析技术研究[D].信息工程大学,2003.
[77]Ker A D. A fusion of maximum likelihood and structural steganalysis [A]. In: Proceedingsof9th International Workshop on Information Hiding [C],2007, LNCS, vol.4567,204-219.
[78]Ker A D. Steganalysis of embedding in two least-significant bits [J]. IEEE Transactions onon Information Forensics and Security,2007,2(1):46-54.
[79]Yang C, Liu F, Luo X, Liu B. Steganalysis frameworks of embedding in multiple least-significant bits [J]. IEEE Transactions on Information Forensics and Security,2008,3(4):662-672.
[80]Yang C, Luo X, Liu F. Embedding ratio estimating for each bit plane of image [A]. In:Proceedings of11th International Workshop on Information Hiding [C],2009, LNCS, vol.5806,59-72.
[81]罗向阳,刘粉林,杨春芳,廉士国.一类自适应隐写的更改比率估计[J].中国科学:信息科学,2011,41(3):297-310.
[82]Luo X, Liu F, Yang C. Steganalysis of adaptive image steganography in multiple Gray codebitplanes [J]. Multimeida Tools and Applications,2012,57(3):651-667.
[83]Luo X, Yang C, Wang D, Liu F. LTSB steganalysis based on quartic equation [J]. LNCSTransactions on Data Hiding and Multimedia Security II,2007,68-90.
[84]Yu X, Noboru B. A fast and effective method to detect multiple least significant bitssteganography [A]. In: Proceedings of2008ACM symposium on Applied computing [C],2008,1443-1447.
[85]Yu X, Tan T, Wang Y. Extended optimization method of LSB steganalysis [A]. In:Proceedings of IEEE International Conference on Image Processing [C],2005,1102-1105.
[86]Yu X, Babaguchi N. Weighted stego-image based steganalysis in multiple least significantbits [A]. In: Proceedings of International Conference on Multimedia and Expo [C],2008,265-268.
[87]B hme R. Weighted stego-image steganalysis for JPEG covers [A]. In: Proceedings of10thInternational Workshop on Information Hiding [C],2008, LNCS, vol.5284,178-194.
[88]Fridrich J, Goljan M, Hogea D. Attacking the Outguess [A]. In: Proceedings of ACMWorkshop on Multimedia and Security [C],2002,967-982.
[89]Mallat S G. A theory for multiresolution signal decomposition: the wavelet representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1989,11(7):674-693
[90]孔祥维,李玲玲,尤新刚.基于多特征支持向量回归的F5定量隐密分析研究[J].计算机研究与发展,2009,46(Suppl. I):157-161.
[91]Smola J, Sch elkopf B. A tutorial on support vector regression [R]. Technical report, NeuroCOLT2Technical Report NC2-TR-1998-030,1998.
[92]Pevny T, Fridrich J. Merging markov and DCT features for multi-class JPEG steganalysis
[A]. In: Proceedings SPIE, Electronic Imaging, Security, Steganography, and Watermarkingof Multimedia Contents IX [C],2007, SPIE, vol.6505,3-1-3-14.
[93]Ker A D. Optimally weighted least-squares steganalysis [A]. In: Proceedings of SPIESecurity, Steganography and Watermarking of Multimedia Contents IX [C],2007, SPIE, vol.6505:601-616.
[94]Westfeld A. Detecting low embedding rates [A]. In: Proceedings of5th InternationalWorkshop on Information Hiding [C],2002, LNCS, vol.2578,324-339.
[95]Westfeld A, Pfitzmann A. Attacks on steganographic systems [A]. In: Proceedings of3rdInternational Workshop on Information Hiding [C],2000, LNCS, vol.1768,61-75.
[96]陈嘉勇,刘九芬,祝跃飞,张卫明.对DCT域连续LSB隐写术的提取攻击.模式识别与人工智能,2011,24(4):484-491.
[97]Ker A D. A weighted stego image detector for sequential LSB replacement LSBreplacement [A]. In: Proceedings of3rd International Symposium on Information Assuranceand Security [C],2007,453-456.
[98]Ker A D, Lubenko I. Feature reduction and payload location with WAM steganalysis [A]. In:Proceedings of SPIE, Media Forensics and Security [C],2009, SPIE, vol.7254,72540A-1-72540A-13.
[99]Quach T T. On locating steganographic payload using residuals A]. In: Proceedings of SPIE,Media Watermarking, Security, and Forensics III [C],2011, SPIE, vol.7880,78800J01-78800J07.
[100] Quach T T. Optimal cover estimation methods and steganographic payload location [J].IEEE Transactions on Information Forensics and Security,2011,6(4):1214-1222.
[101]周治平,王永志,纪志成等.隐写路径搜寻方法研究[J].哈尔冰工业大学学报,2006,38(S):768-772.
[102]陈嘉勇,祝跃飞,张卫明,刘九芬.对随机LSB隐写术的选择密钥提取攻击[J].通信学报,2010,31(5):73-80.
[103]苏育才,姜翠波,张跃辉.矩阵理论[M].北京:科学出版社,2006.
[104] B hme R, Ker A D. A two-factor error model for quantitative steganalysis [A]. In:Proceedings of SPIE, Security, Steganography, and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol. vol.6072,59-74.
[105] Ker A D. Derivation of Error Distribution in Least Squares Steganalysis [J]. IEEETransactions on Information Forensics and Security,2007,2(2):140-148.
[106] Cover T M, Thomas J A. Elements of Information Theory [M]. Beijing: TsinghuaUniversity Press,2003.
[107] Han J, Kamber M著.范明,孟小峰译.数据挖掘——概念与技术[M].机械工业出版社,2006:204.
[108] Fridrich J. Feature-based steganalysis for JPEG images and its implications for futuredesign of steganographic schemes [A]. In: Proceeding of6th International Workshop onInformation Hiding,2004, LNCS, vol.3200,67-81.
[2] Herodotus. The Histories [M]. London, England: J. M. Dent&Sons, Ltd,1992.
[3] Kurak C, McHugh J. A cautionary note on image downgrading [A]. In: Proceedings of the8th IEEE Annual Computer Security Applications Conference [C],1992,153-159.
[4] Simmons G J. The prisoners’ problem and the subliminal channel [A]. In: Proceedings ofCRYPTO [C],1983,51-67.
[5] Cachin C. An information-theoretic model for steganography [A]. In: Proceedings of2thInternational Workshop on Information Hiding [C],1998, LNCS, vol.1525,306-318.
[6] Cachin C. An information-theoretic model for steganography [J]. Information andComputation,2004,192(1):41-56.
[7] Hopper N J, Langford J, Ahn L V. Provably secure steganography [A]. In: Proceedings pfCRYPTO [C],2002, LNCS, vol.2442,77-92.
[8]刘光杰,戴跃伟,赵玉鑫,王执铨.隐写对抗的博弈论建模[J].南京理工大学(自然科学版),2008,32(2):199-204.
[9] Fridrich J, Goljan M. Digital image steganography using stochastic modulation [A]. In:Proceedings of SPIE, Security and Watermarking of Multimedia Contents V [C].2003,SPIE, vol.5020,191-202.
[10]Yang C, Liu F, S Lian, Luo X. Weighted stego-image steganalysis of message hidden intoeach bit plane [J]. The Computer Journal,2011, doi:10.1093/comjnl/bxr112.
[11]Fridrich J, Soukal D, Goljan M. Maximum likelihood estimation of length of secret messageembedded using+-K steganography in spatial domain [A]. In: Proceedings of SPIE,Electronic Imaging, Security, Steganography and Watermarking of Multimedia Contents VII
[C],2005, SPIE vol.5681,595-606.
[12]Fridrich J, Rui D. Secure steganographic methods for palette images [A]. In: Proceedings of3rd International Workshop on Information Hiding [C], LNCS, vol.1768,2000,47-60.
[13]Pan H K, Chen Y Y, Tseng Y C. A secure data hiding scheme for two color images [A]. In:Proceedings of IEEE Symposium on Computer and Communications [C],2000,750-755.
[14]Tseng Y C, Pan H K. Data hiding in2-color images [J]. IEEE Transactions on Computers.2002,51(7):873-878.
[15]Kawaguchi E, Eason R O. Principle and applications of BPCS-steganography [A]. In:Proceedings of SPIE, Multimedia Systems and Applications [C],1998, SPIE vol.3528,464-473.
[16]Nguyen B, Yoon S, Lee H. Multi bit plane image steganography [A]. In: Proceeding ofInternational Workshop on Digital Watermarking [C],2006, LNCS, vol.4283,61-70.
[17]Luo W, Huang F, Huang J. Edge adative image steganography based on LSB matchingrevisited [J]. IEEE Transactions on Information Forensics and Security,5(2):201-214.
[18]Katzenbeisser S, Petitcolas F A P编.吴秋新,钮心忻,杨义先等译.信息隐藏技术—隐写术与数字水印[M].北京:人民邮电出版社,2001.
[19]Fridrich J, Pevny. Statistically undetectable JPEG steganography: dead ends challenges, andopportunities [A]. In: Proceedings of9th Workshop on Multimedia&Security [C],2007,3-14.
[20]Provos N. Defending against statistical steganalysis [A]. In: Proceedings of10th USENIXSecurity Symposium [C],2001,323-335.
[21]Hetzl S, Mutzel. A graph-theoretic approach to steganography [A]. In: Proceedings of9thIFIP TC-6TC-11International Conference on Communications and Multimedia Security,Lecture Notes in Computer Science [C],2005, vol.3677,119-128.
[22]Sallee P. Model-based steganography [A]. In: Proceedings of International Workshop onDigital Watermarking [C],2003, LNCS, vol.2939,154-167.
[23]Sallee P. Model-based methods for steganography and steganalysis [J]. International Journalof Image Graphics,2005,5(1):167-190.
[24]Westfeld A. High capacity despite better steganalysis (F5-a steganographic algorithm)[A].In: Proceedings of4th International Workshop on Information Hiding [C],2001, LNCS, vol.2137,289-302.
[25]Kim Y, Duric Z, Richards D. Modified matrix encoding technique for minimal distortionsteganography. Information Hiding [A]. In: Proceedings of8th International Workshop onInformation Hiding [C],2006, LNCS, vol.4437,314-327.
[26]Fridrich J, Goljan M, Soukal D. Perturbed quantization steganography [J]. ACM MultimediaSystem Journal,2005,11(2):98-107.
[27]Low S H, Maxemchuk N F, Lapone A M. Document identification for copyright protectionusing centroid detection [J]. IEEE Transaction on Communications,1998,46(3):372-383.
[28]刘东,孙明,周明天.基于图论的文本数字水印技术[J].计算机研究与发展,2007,44(10):1757-1764.
[29]Wayner P. Mimic Functions [J]. Cryptologia,1992, XVI(3):193-214.
[30]Wayner P. Strong theoretical steganography [J]. Cryptologia,1995, XIX(3):285-299.
[31]Fridrich J, Goljan M. Practical steganalysis of digital images—state of the art [A]. In:Proceedings of SPIE, Security and Watermarking of Multimedia Contents IV [C],2002,SPIE, vol.4675,1-13.
[32]钮心忻.信息隐藏与数字水印[M].北京:北京邮电大学出版社,2004.
[33]Luo X, Wang D, Wang P, Liu F. A review on blind detection for image steganography [J].Signal Processing,2008,88(9):2138-2157.
[34]Avcibas, Memon N, Sankur B. Steganalysis using image quality metrics [J]. IEEETransactions on Image Processing,2003,12(2):221-229.
[35]Farid H. Detecting steganographic messages in digital images [R]. Hanover, NH: DartmouthCollege,2001.
[36]Goljan M, Fridrich J, Holotyak T. New blind steganalysis and its implications [A]. In:Proceedings of SPIE, Security, Steganography, and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol.6072,1-13.
[37]Harmsen J J, Pearlman W A. Steganalysis of additive noise modelable information hiding
[A]. In: Proceedings of SPIE, Security, Steganography, and Watermarking of MultimediaContents V [C],2003, SPIE, vol.5020,131-142.
[38]Shi Y, Xuan G, Yang C, Gao J, and et al. Effective steganalysis based on statistical momentsof wavelet characteristic function [A]. In: Proceedings of IEEE International Conference onInformation Technology: Coding and Computing [C],2005,768-773.
[39]Chen X C, Wang Y H, Tan T N, Guo L. Blind image steganalysis based on statisticalanalysis of empirical matrix [A]. In: Proceedings of18th International Conference onPattern Recognition [C],2006, vol.3,1107-1110.
[40]Lie W N, Lin G S. A feature-based classification technique for blind image steganalysis [J].IEEE Transactions on Multimedia,2005,7(6):1007-1020.
[41]Pevny T, Fridrich J. Towards multi-class blind steganalyzer for JPEG images [A]. In:Proceedings of4th International Workshop on Digital Watermarking [C],2005, LNCS, vol.3710,39-53.
[42]Pevny T, Fridrieh J. Multiclass detector of current steganographic methods for JPEG format[J]. IEEE Transactions on Information Forensics and Security,2008,3(4):635-650.
[43]Dong J, Wang W, Tan T N. Multi-class blind steganalysis based on image run-lengthanalysis. In: Proceedings of7th International Workshop on Digital Watermarking [C],2009,LNCS, vol.5703,199-210.
[44]Chiew K L, Josef P. Binary Image steganographic techniques classification based on multi-class steganalysis [A]. In: Proceedings of6th International Conference on InformationSecurity, Practice and Experience [C],2010, LNCS, vol.6047,341-358.
[45]Fridrich J, Goljan M, Hogea D, Souka D. Quantitative steganalysis of digital images:estimating the secret message length [J]. ACM Multimedia Systems Journal, Special issueon Multimedia Security,2003,9(3):288-302.
[46]Dabeer O, Sullian K, Madhow U, et al. Detection of hiding in the least significant bit [J].IEEE Transactions on Signal Processing,2004,52(10):3046-3058.
[47]Fridrich J, Goljan M, Hogea D. Steganalysis of JPEG images: breaking the F5algorithm [A].In: Proceedings of5th International Workshop on Information Hiding [C],2002, LNCS, vol.258,310-323.
[48]Fridrich J, Goljan M, Du R. Reliable detection of LSB steganography in color and grayscaleimages [A]. In: Proceedings of Workshop on Multimedia&Security [C],2001,27-30.
[49]Dumitrescu S, Wu X, Wang Z. Detection of LSB steganography via sample pair analysis [J].IEEE Transactions on Signal Processing,2003,51(7):1995-2007.
[50]Fridrich J, Goljan M. On estimation of secret message length in LSB steganography inspatial domain [A]. In: Proceedings of SPIE Security, Steganography and Watermarking ofMultimedia Contents VI [C],2004, SPIE, vol.5306,23-24.
[51]Ker A D, B hme R. Revisiting weighted stego-image steganalysis [A]. In: Proceedings ofSPIE Security, Steganography and Watermarking of Multimedia Contents X [C],2007,SPIE, vol.6819,27-31.
[52]Fridrich J, Soukal D, Goljan M. Maximum likelihood estimation of length of secret messageembedded using+-K steganography in spatial domain [A]. In: Proceedings of SPIE,Electronic Imaging, Security, Steganography and Watermarking of Multimedia Contents VII
[C],2005, SPIE, vol.5681,595-606.
[53]He J, Huang J. Steganalysis of stochastic modulation steganography [J]. Science in China:Series F Information Sciences,2006,49(3):273-285.
[54]Pevny T, Fridrich J, Ker A D. From blind to quantitative steganalysis [A]. In: Proceedings ofSPIE, Electronic Imaging, Security and Forensics of Multimedia Contents XI [C],2009,SPIE, vol.7254,0C-1-0C-14.
[55]Guo Y, Kong X, You X, Li L. Universal methodology for developing quantitativesteganalysis [J]. Chinese Journal of Electronics,2009,18(3):455-459.
[56]Trivedi S, Chandramouli R. Secret key estimation in sequential steganography [J]. IEEETransactions on Signal Processing,2005,53(2):746-757.
[57]Kong X, Liu W, You X. Secret message location steganalysis based on local coherences ofhue [A]. In: Proceeding of Pacific Rim Conference on Multimedia Part II [C],2005, LNCS,vol.3768,301-311.
[58]Ambalavanan A, Chandramouli R. A Bayesian image steganalysis approach to estimate theembedded secret message [A]. In: Proceedings of7th ACM Workshop on Multimedia andSecurity [C],2005,33-38.
[59]Davidson I, Paul G. Locating secret messages in images [A]. In: Proceedings of10th ACMSIGKDD International Conference on Knowledge Discovery and Data Mining [C],2004,545-550.
[60]Ker A D. Locating steganographic payload via WS residuals [A]. In: Proceedings of10thWorkshop on Multimedia and Security [C],2008,27-31.
[61]Chiew K L, Pieprzyk J. Identifying steganographic payload location in binary image [A]. In:Proceeding of Pacific Rim Conference on Multimedia Part I [C],2010, LNCS, vol.6297,590-600.
[62]Provos N, Honeyman P. Detecting steganographic content on the Internet [R]. InternetSociety, Network and Distributed System Security Symposium,2002.
[63]张卫明,李世取,刘九芬.隐写术CPT的等价密钥分析[J].电子学报,2007,35(12):2258-2261.
[64]Fridrich J, Goljan M, Soukal D. Searching for the stego-key [A]. In: Proceedings of SPIESecurity, Steganography, and Watermarking of Multimedia Contents VII [C],2004, SPIE,vol.5306,70-82.
[65]Fridrich J, Goljan M, Soukal D, Holotyak T. Forensic steganalysis: determining the stegokey in spatial domain steganography [A]. In: Proceedings of SPIE Security, Steganography,and Watermarking of Multimedia Contents VII [C],2005, SPIE vol.5681,631-642.
[66]张卫明,李世取,刘九芬.对空域图像LSB隐写术的提取攻击[J].计算机学报,2007，30(9):1625-1631.
[67]张卫明.隐写信息提取的理论与方法研究[D].信息工程大学,2005.
[68]张卫明,刘九芬,李世取. LSB隐写术的密钥恢复方法[J].中山大学学报(自然科学版),2005,44(3):29-33.
[69]Zhang T, Ping X J. A fast and effective steganalytic technique against Jsteg-like algorithms
[A]. In: Proceedings of Symposium on Applied Computing [C],2003,307-311.
[70]Zhang T, Ping X J. A new approach to reliable detection of LSB steganography in naturalimage [J]. Signal Process,2003,83(10):545-548.
[71]Lu P, Luo X, Tang Q, Zou Y. An improved sample pairs method for detection of LSBembedding [A]. In: Proceedings of6th Workshop on Information Hiding [C],2004, LNCS,vol.3200,116-127.
[72]Ker A D. A general framework for the structural steganalysis of LSB replacement [A]. In:Proceedings of7th International Workshop on Information Hiding [C],2005, LNCS, vol.3727:296-311.
[73]Ker A D. Fourth-order structural steganalysis and analysis of cover assumptions [A]. In:Proceedings of SPIE, Security, Steganography and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol.6072:25-38.
[74]Dumitrescu S, Wu X. A new framework of LSB steganalysis of digital media [J]. IEEETransactions on Signal Process,2005,53(10):3936-3947.
[75]Luo X, Yang C, Liu F. Equivalence analysis among DIH, SPA, and RS steganalysismethods [A]. In: Proceedings of the10th IFIP conference of Communication andMultimedia Security [C],2006, LNCS, vol.4237,161-172.
[76]张涛.图象隐写分析技术研究[D].信息工程大学,2003.
[77]Ker A D. A fusion of maximum likelihood and structural steganalysis [A]. In: Proceedingsof9th International Workshop on Information Hiding [C],2007, LNCS, vol.4567,204-219.
[78]Ker A D. Steganalysis of embedding in two least-significant bits [J]. IEEE Transactions onon Information Forensics and Security,2007,2(1):46-54.
[79]Yang C, Liu F, Luo X, Liu B. Steganalysis frameworks of embedding in multiple least-significant bits [J]. IEEE Transactions on Information Forensics and Security,2008,3(4):662-672.
[80]Yang C, Luo X, Liu F. Embedding ratio estimating for each bit plane of image [A]. In:Proceedings of11th International Workshop on Information Hiding [C],2009, LNCS, vol.5806,59-72.
[81]罗向阳,刘粉林,杨春芳,廉士国.一类自适应隐写的更改比率估计[J].中国科学:信息科学,2011,41(3):297-310.
[82]Luo X, Liu F, Yang C. Steganalysis of adaptive image steganography in multiple Gray codebitplanes [J]. Multimeida Tools and Applications,2012,57(3):651-667.
[83]Luo X, Yang C, Wang D, Liu F. LTSB steganalysis based on quartic equation [J]. LNCSTransactions on Data Hiding and Multimedia Security II,2007,68-90.
[84]Yu X, Noboru B. A fast and effective method to detect multiple least significant bitssteganography [A]. In: Proceedings of2008ACM symposium on Applied computing [C],2008,1443-1447.
[85]Yu X, Tan T, Wang Y. Extended optimization method of LSB steganalysis [A]. In:Proceedings of IEEE International Conference on Image Processing [C],2005,1102-1105.
[86]Yu X, Babaguchi N. Weighted stego-image based steganalysis in multiple least significantbits [A]. In: Proceedings of International Conference on Multimedia and Expo [C],2008,265-268.
[87]B hme R. Weighted stego-image steganalysis for JPEG covers [A]. In: Proceedings of10thInternational Workshop on Information Hiding [C],2008, LNCS, vol.5284,178-194.
[88]Fridrich J, Goljan M, Hogea D. Attacking the Outguess [A]. In: Proceedings of ACMWorkshop on Multimedia and Security [C],2002,967-982.
[89]Mallat S G. A theory for multiresolution signal decomposition: the wavelet representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1989,11(7):674-693
[90]孔祥维,李玲玲,尤新刚.基于多特征支持向量回归的F5定量隐密分析研究[J].计算机研究与发展,2009,46(Suppl. I):157-161.
[91]Smola J, Sch elkopf B. A tutorial on support vector regression [R]. Technical report, NeuroCOLT2Technical Report NC2-TR-1998-030,1998.
[92]Pevny T, Fridrich J. Merging markov and DCT features for multi-class JPEG steganalysis
[A]. In: Proceedings SPIE, Electronic Imaging, Security, Steganography, and Watermarkingof Multimedia Contents IX [C],2007, SPIE, vol.6505,3-1-3-14.
[93]Ker A D. Optimally weighted least-squares steganalysis [A]. In: Proceedings of SPIESecurity, Steganography and Watermarking of Multimedia Contents IX [C],2007, SPIE, vol.6505:601-616.
[94]Westfeld A. Detecting low embedding rates [A]. In: Proceedings of5th InternationalWorkshop on Information Hiding [C],2002, LNCS, vol.2578,324-339.
[95]Westfeld A, Pfitzmann A. Attacks on steganographic systems [A]. In: Proceedings of3rdInternational Workshop on Information Hiding [C],2000, LNCS, vol.1768,61-75.
[96]陈嘉勇,刘九芬,祝跃飞,张卫明.对DCT域连续LSB隐写术的提取攻击.模式识别与人工智能,2011,24(4):484-491.
[97]Ker A D. A weighted stego image detector for sequential LSB replacement LSBreplacement [A]. In: Proceedings of3rd International Symposium on Information Assuranceand Security [C],2007,453-456.
[98]Ker A D, Lubenko I. Feature reduction and payload location with WAM steganalysis [A]. In:Proceedings of SPIE, Media Forensics and Security [C],2009, SPIE, vol.7254,72540A-1-72540A-13.
[99]Quach T T. On locating steganographic payload using residuals A]. In: Proceedings of SPIE,Media Watermarking, Security, and Forensics III [C],2011, SPIE, vol.7880,78800J01-78800J07.
[100] Quach T T. Optimal cover estimation methods and steganographic payload location [J].IEEE Transactions on Information Forensics and Security,2011,6(4):1214-1222.
[101]周治平,王永志,纪志成等.隐写路径搜寻方法研究[J].哈尔冰工业大学学报,2006,38(S):768-772.
[102]陈嘉勇,祝跃飞,张卫明,刘九芬.对随机LSB隐写术的选择密钥提取攻击[J].通信学报,2010,31(5):73-80.
[103]苏育才,姜翠波,张跃辉.矩阵理论[M].北京:科学出版社,2006.
[104] B hme R, Ker A D. A two-factor error model for quantitative steganalysis [A]. In:Proceedings of SPIE, Security, Steganography, and Watermarking of Multimedia ContentsVIII [C],2006, SPIE, vol. vol.6072,59-74.
[105] Ker A D. Derivation of Error Distribution in Least Squares Steganalysis [J]. IEEETransactions on Information Forensics and Security,2007,2(2):140-148.
[106] Cover T M, Thomas J A. Elements of Information Theory [M]. Beijing: TsinghuaUniversity Press,2003.
[107] Han J, Kamber M著.范明,孟小峰译.数据挖掘——概念与技术[M].机械工业出版社,2006:204.
[108] Fridrich J. Feature-based steganalysis for JPEG images and its implications for futuredesign of steganographic schemes [A]. In: Proceeding of6th International Workshop onInformation Hiding,2004, LNCS, vol.3200,67-81.