针对图像Two-LSBs隐写的分析方法

针对图像Two-LSBs隐写算法,提出了一种新的隐写分析方法.该方法根据Two-LSBs隐写对图像像素相关性的改变来判断秘密信息的有无,同时还可以准确地估计隐写嵌入率和秘密信息的长度.实验结果表明该分析方法的有效性,当嵌入率高于20%时检测正确率可达90%.     

应用共生矩阵特征的改进自适应JPEG隐写

针对传统图像隐写算法抵抗通用盲检测能力不足的问题,提出一种基于共生矩阵特征的自适应JPEG隐写算法。该算法在分析图像直方图特征和分块效应特征的基础上,引入能够反映共生矩阵特征的块间系数相关性设计了失真函数,并结合网格码（STC）完成秘密信息的嵌入。实验结果表明,面对三种通用隐写检测算法,算法较以往隐写算法具有更强的安全性和抗检测普适性。     

针对MP3Stego的一种边信息作为特征的改进分析算法

针对利用MP3边信息中主数据起始位置值作为特征的隐写分析方法存在嵌入少量秘密信息时检测率低的问题进行了改进,将MP3Stego隐写前后的数据帧主数据起始位置差扩大,以减小特征提取时重压缩估计方法带来的误差,从而提高了音频隐写分析检测率.实验结果表明,该算法对MP3Stego音频隐写可进行有效的隐写分析与检测,尤其是在嵌入少量秘密信息时检测率较现有算法有明显提高.     

基于MP3的后置式自适应隐写算法

针对MP3隐写算法较为复杂、隐写容量较小的问题,提出一种基于MP3的后置式隐写算法。该算法根据待嵌入秘密信息的多少以及采用Hb码表编码的小值区码字数目的多少,自适应、较均匀且随机地选取嵌入位置,通过改变载体码字的末位实现秘密信息的嵌入。提取信息时无需解码,找到嵌入位置,提取码字的末位从而恢复秘密信息。所提算法在容量方面较MP3stego算法提升约3倍以上,在不可感知性方面较以往采用码字映射的隐写算法提升约10%;另外秘密信息提取时无需解码,大大提高了效率。实验证明了所提算法的安全性、时效性,其能基本满足安全通信的需要。     

结合失真函数的稀疏域图像隐写算法

稀疏域图像隐写利用稀疏表示在载体图像中嵌入秘密信息。提出设计稀疏域图像隐写的失真函数,通过选择使失真函数最小的修改方式进行秘密嵌入,从而减小秘密信息嵌入对载体的影响。实验结果表明,该算法可以在保持图像视觉效果的同时,更有效地抵抗现有的隐写分析算法。     

基于小波域隐马尔可夫树模型和SVR的定量隐写分析

本文结合小波域隐马尔可夫树模型(Hidden Markov Tree,HMT)和支持向量回归(Support Vector Regression,SVR)分析法提出了估计信息嵌入率的定量隐写分析。本文分别将垂直、水平和对角的小波系数建模为两状态的HMT模型,用EM算法(Estimation and Maximization Algorithm)对HMT模型进行训练得出包括状态初始概率、小波子带的转移概率和方差在内的模型参数,由这些参数构成图像的特征。然后结合SVR建立定量隐写分析器,并估计F5,outguess和MB三种隐写图的信息嵌入率。     

分级安全的文本隐写方法

针对单一数据类型隐写方法安全性不高、隐写容量不足等问题,提出了一种具有分级安全的文本隐写方法。首先,将整个载体文档中的多种类型的数据作为备选隐写载体,以不同类型数据的隐写特点和隐写分析技术为评估依据定义隐写安全等级,构建了一个多类数据融合的分级安全隐写模型。然后,根据秘密信息的长度自适应确定安全等级,并利用分级安全隐写模型将秘密信息分块地嵌入在同一个载体文档相互独立的多个不同类数据中。理论分析及实验结果表明,与现有基于单一数据类型隐写方法相比,所提方法扩大了隐写容量,在嵌入等量的秘密信息情况下,降低了文档中同一类载体数据的统计特征改变程度,提高了秘密信息的整体安全性。     

基于运动矢量直方图不变的数字视频隐写算法

针对现有基于运动矢量(MV)的数字视频隐写算法对载体直方图统计特性带来较大改变这一问题,提出一种基于运动矢量直方图不变的数字视频隐写算法。利用保持直方图数据映射方法,把秘密信息隐藏在视频运动矢量中;同时,利用数据匹配编码对嵌入之前的秘密信息进行编码处理,得到了与视频运动矢量统计特征基本匹配的数据流,使算法接近于信息论下的完美安全。实验结果表明：算法有效控制了运动矢量直方图的改变,同时码率的增加也被有效地控制在1%以内,隐写分析检出率平均下降了30%~50%。     

结合视觉感知特性和湿纸编码的视频隐写算法

利用双树复小波变换提取出视频图像的运动特征图、对比敏感图、纹理特征图和场景变化情况,综合确定视觉不敏感点,从而构造出用于嵌入秘密信息的隐写位置图,并结合湿纸编码实现了信息的隐写。算法不需要接收者知道具体的嵌入位置便能够实现信息的提取,避免了重复获得隐写位置图的大量运算,并且不必考虑增加冗余以修正隐写操作引起的隐写位置重新判断误差,提高了提取效率和安全性。实验表明,算法具有较高的嵌入容量和良好的隐写不可见性,可以满足隐蔽通信的要求。     

彩色图像通用隐写分析的多类统计特征

提出了用于彩色图像通用隐写分析的3类统计特征。针对隐写所导致的图像DCT系数分布、空域像素值平坦性以及颜色空间一致性的改变,从系数分布模型背离程度、空域相关性和不同颜色分量间的相关性3个方面设计能够体现原始载体和含密图像差异的特征,提取了10维特征向量。这些特征对隐写行为较为敏感、涵盖面广,且维数低、计算方便。在这些特征的基础上,可利用支持向量机分析彩色JPEG图像是否含有秘密信息。实验结果表明,基于这些特征的通用隐写分析方法可以准确、高效地检测以Jstge、F5或MB隐写方法嵌入的秘密信息,辨别隐写方法种类,具有很好的通用性。     

Quantitative steganalysis of digital images: estimating the secret message length

The objective of steganalysis is to detect messages hidden in cover objects, such as digital images. In practice, the steganalyst is frequently interested in more than whether or not a secret message is present. The ultimate goal is to extract and decipher the secret message. However, in the absence of the knowledge of the stego technique and the stego and cipher keys, this task may be extremely time consuming or completely infeasible. Therefore, any additional information, such as the message length or its approximate placement in image features, could prove very valuable to the analyst. In this paper, we present general principles for developing steganalytic methods that can accurately estimate the number of changes to the cover image imposed during embedding. Using those principles, we show how to estimate the secret message length for the most common embedding archetypes, including the F5 and OutGuess algorithms for JPEG, EzStego algorithm with random straddling for palette images, and the classical LSB embedding with random straddling for uncompressed image formats. The paper concludes with an outline of ideas for future research such as estimating the steganographic capacity of embedding algorithms.     

Steganalysis Frameworks of Embedding in Multiple Least-Significant Bits

Replacement of least-significant bit plane is one of the popular steganography techniques in digital images because of its extreme simplicity. But it is more difficult to precisely estimate the rate of secret message embedded by replacement of multiple least-significant bit (MLSB) planes of a carrier object. In order to model the MLSB embedding, a lemma is introduced to prove the transition relationships among some trace subsets. Then, based on these transition relationships, two novel steganalysis frameworks are designed to detect two kinds of distinct MLSB embedding methods. A series of experiments show that the proposed steganalysis frameworks are highly sensitive to MLSB steganography, and can estimate the rate of secret message with higher accuracy. Furthermore, these frameworks can fully meet the need to distinguish stego images under low false positive rate, especially when the embedded message is short.      

Improved universal quantitative steganalysis in spatial domain using ELM ensemble

Quantitative steganalysis seeks to extract the additional information about the hidden message in the covert communications. Most of the quantitative steganalyzers available in the literature target a specific embedding algorithm and generally extract the payload information using structural paradigm. Modern steganalyzers use supervised machine learning to estimate the stego payload using sophisticated feature sets. In this paper, an Ensemble Framework based universal quantitative steganalyzer for digital images is proposed which employs optimised Extreme Learning Machines as the base regressors. The framework exploits inherent diversity of the base regressor and the use of random subspaces of the image features further reduces the prediction error. The proposed ensemble regressor exhibits improved payload predictions when evaluated vis-à-vis the individual base regressor and other state-of-the-art algorithms. The experimental results across different embedding algorithms, image datasets and variedly sized feature sets demonstrate the robustness and wide applicability of the proposed framework.

     

A novel steganalysis for TPVD steganographic method based on differences of pixel difference histogram

Tri-way Pixel Value Differencing (TPVD) steganographic method is a new modified version of another well-known method called PVD, which intents to increase embedding capacity and security of its successor by hiding secret bits in both vertical and diagonal edges of a cover image, in addition to the horizontal edges used in PVD. In this paper, it is shown that the histogram of difference values of a stego image under the TPVD algorithms is vulnerable to a particular statistical analysis. So, a new steganalytic measure named Growing Anomalies is introduced that its value has a linear relationship with secret message rate. It is shown empirically and theoretically that proposed steganalysis method based on this measure can estimate the amount of secret bits with a negligible error rate. The proposed steganalyser can classify test images as stego or cover with 97% accuracy when they contain more that 10% secret data. Implementation results indicate that proposed method can estimate secret message rate with an average accuracy of 95%.     

Steganalysis of stochastic modulation steganography

Steganography[1] conceals the occurrence and real purpose of communication by em- bedding message into the cover medium, such as an image, audio or video. Because im- age is widely used in Internet and has much redundancy, it has been one of the most im- …     

基于邻域相关性MC模型的空域隐写检测算法

空域隐写作为最常用的隐写方式之一,对其进行隐写分析具有较高的实用价值。空域隐写破坏了图像的邻域相关性,通过分析其在隐写前后发生的改变,能有效检测图像是否进行了隐写。如何对邻域相关性进行量化分析成为问题解决的关键。通过对图像邻域相关性进行MC建模,提出了一种有效的空域隐写检测算法。该算法在相关性MC模型的基础上提取出相关性量化矩阵,分析了其在隐写前后发生的变化,并提取出相应的检测特征。最后,以Fisher分类器作为检测分类器进行了检测实验。实验结果表明该算法能有效检测多种空域隐写算法,其平均检测率达到85％以上。与其它空域隐写检测算法相比该算法具有更好的适用性。     

生成对抗网络驱动的图像隐写与水印模型

目的 图像信息隐藏包括图像隐写术和图像水印技术两个分支。隐写术是一种将秘密信息隐藏在载体中的技术,目的是为了实现隐秘通信,其主要评价指标是抵御隐写分析的能力。水印技术与隐写术原理类似,但其是通过把水印信息嵌入到载体中以达到保护知识产权的作用,追求的是防止水印被破坏而尽可能地提高水印信息的鲁棒性。研究者们试图利用生成对抗网络（generative adversarial networks,GANs）进行自动化的隐写算法以及鲁棒水印算法的设计,但所设计的算法在信息提取准确率、嵌入容量和隐写安全性或水印鲁棒性、水印图像质量等方面存在不足。方法 本文提出了基于生成对抗网络的新型端到端隐写模型（image information hiding-GAN,IIH-GAN）和鲁棒盲水印模型（image robust blind watermark-GAN,IRBW-GAN）,分别用于图像隐写术和图像鲁棒盲水印。网络模型中使用了更有效的编码器和解码器结构SE-ResNet（squeeze and excitation ResNet）,该模块根据通道之间的相互依赖性来自适应地重新校准通道方式的特征响应。结果 实验结果表明隐写模型IIH-GAN相对其他方法在性能方面具有较大改善,当已知训练好的隐写分析模型的内部参数时,将对抗样本加入到IIH-GAN的训练过程,最终可以使隐写分析模型的检测准确率从97.43%降低至49.29%。该隐写模型还可以在256×256像素的图像上做到高达1 bit/像素（bits-per-pixel）的相对嵌入容量;IRBW-GAN水印模型在提升水印嵌入容量的同时显著提升了水印图像质量以及水印提取正确率,在JEPG压缩的攻击下较对比方法提取准确率提高了约20%。结论 本文所提IIH-GAN和IRBW-GAN模型在图像隐写和图像水印领域分别实现了领先于对比模型的性能。     

Efficient steganalysis of images: learning is good for anticipation

This paper describes a Bayesian formalism for digital image steganalysis allowing the detection of stego images, the identification of the steganographic algorithm used, the estimation of message length and location, and anticipation in the case of embedding using an unknown steganographic algorithm. A Bayesian multinomial logistic regression based on a variational approximation is proposed. Detection, identification, and anticipation involve discriminative learning in feature space. Estimation requires the fusion of classifiers allowing discrimination between fully embedded and cover subimages of different sizes. The validation on JPEG images shows that the proposed scheme is effective and allows the anticipation of unknown steganographic algorithms.     

基于注意力机制的高容量通用图像隐写模型

随着深度学习与隐写技术的发展,深度神经网络在图像隐写领域的应用越发广泛,尤其是图像嵌入图像这一新兴的研究方向.主流的基于深度神经网络的图像嵌入图像隐写方法需要将载体图像和秘密图像一起输入隐写模型生成含密图像,而最近的研究表明,隐写模型仅需要秘密图像作为输入,然后将模型输出的含密扰动添加到载体图像上,即可完成秘密图像的嵌入过程.这种不依赖载体图像的嵌入方式极大地扩展了隐写的应用场景,实现了隐写的通用性.但这种嵌入方式目前仅验证了秘密图像嵌入和恢复的可行性,而对隐写更重要的评价标准,即隐蔽性,未进行考虑和验证.提出一种基于注意力机制的高容量通用图像隐写模型USGAN,利用注意力模块, USGAN的编码器可以在通道维度上对秘密图像中像素位置的扰动强度分布进行调整,从而减小含密扰动对载体图像的影响.此外,利用基于CNN的隐写分析模型作为USGAN的目标模型,通过与目标模型进行对抗训练促使编码器学习生成含密对抗扰动,从而使含密图像同时成为攻击隐写分析模型的对抗样本.实验结果表明,所提模型不仅可以实现不依赖载体图像的通用嵌入方式,还进一步提高了隐写的隐蔽性.     

Simple and effective approach for construction of universal blind steganalyzer

The goal of universal blind steganalysis is to detect all known (already existing) and unknown (previously unseen) steganographic algorithms without knowledge of the exact stego algorithm used by the steganographer. However, a binary blind steganalyzer trained on cover images and stego images randomly selected from “known stego images” (i.e., stego images produced by multiple “known” stego methods with a mixture of payloads), may fail catastrophically on unknown stego methods although shows superior performance on known stego methods. Additionally, unsupervised outlier detection and one-class classification approaches are less likely to fail to detect unknown stego methods but yield high false positive rates. Motivated by these observations, we explore a simple and effective approach for construction of universal blind steganalyzer to achieve overall good performance on both known and unknown stego algorithms. First, we compute Local Outlier Factor (LOF) scores of known stego sample points (feature vectors) with respect to test sample points. Then, we choose stego images with the lowest LOF scores from known stego images as training stego images. Finally, we train a binary classifier on cover images and chosen training stego images for test. Experimental results confirm that the proposed approach performs significantly better than the random sampling-based binary classification method, unsupervised outlier detection and one-class classification approaches on both known and unknown stego algorithms.