A HEVC Video Steganalysis Method Using the Optimality of Motion Vector Prediction

Among steganalysis techniques,detection against MV(motion vector)domain-based video steganography in the HEVC(High Efficiency Video Coding)standard remains a challenging issue.For the purpose of improving the detection performance,this paper proposes a steganalysis method that can perfectly detectMV-based steganography in HEVC.Firstly,we define the local optimality of MVP(Motion Vector Prediction)based on the technology of AMVP(Advanced Motion Vector Prediction).Secondly,we analyze that in HEVC video,message embedding either usingMVP index orMVD(Motion Vector Difference)may destroy the above optimality of MVP.And then,we define the optimal rate of MVP as a steganalysis feature.Finally,we conduct steganalysis detection experiments on two general datasets for three popular steganographymethods and compare the performance with four state-ofthe-art steganalysis methods.The experimental results demonstrate the effectiveness of the proposed feature set.Furthermore,our method stands out for its practical applicability,requiring no model training and exhibiting low computational complexity,making it a viable solution for real-world scenarios.

MI-STEG:A Medical Image Steganalysis Framework Based on Ensemble Deep Learning

Medical image steganography aims to increase data security by concealing patient-personal information as well as diagnostic and therapeutic data in the spatial or frequency domain of radiological images.On the other hand,the discipline of image steganalysis generally provides a classification based on whether an image has hidden data or not.Inspired by previous studies on image steganalysis,this study proposes a deep ensemble learning model for medical image steganalysis to detect malicious hidden data in medical images and develop medical image steganography methods aimed at securing personal information.With this purpose in mind,a dataset containing brain Magnetic Resonance(MR)images of healthy individuals and epileptic patients was built.Spatial Version of the Universal Wavelet Relative Distortion(S-UNIWARD),Highly Undetectable Stego(HUGO),and Minimizing the Power of Optimal Detector(MIPOD)techniques used in spatial image steganalysis were adapted to the problem,and various payloads of confidential data were hidden in medical images.The architectures of medical image steganalysis networks were transferred separately from eleven Dense Convolutional Network(DenseNet),Residual Neural Network(ResNet),and Inception-based models.The steganalysis outputs of these networks were determined by assembling models separately for each spatial embedding method with different payload ratios.The study demonstrated the success of pre-trained ResNet,DenseNet,and Inception models in the cover-stego mismatch scenario for each hiding technique with different payloads.Due to the high detection accuracy achieved,the proposed model has the potential to lead to the development of novel medical image steganography algorithms that existing deep learning-based steganalysis methods cannot detect.The experiments and the evaluations clearly proved this attempt.

Image Steganalysis Based on Deep Content Features Clustering

The training images with obviously different contents to the detected images will make the steganalysis model perform poorly in deep steganalysis.The existing methods try to reduce this effect by discarding some features related to image contents.Inevitably,this should lose much helpful information and cause low detection accuracy.This paper proposes an image steganalysis method based on deep content features clustering to solve this problem.Firstly,the wavelet transform is used to remove the high-frequency noise of the image,and the deep convolutional neural network is used to extract the content features of the low-frequency information of the image.Then,the extracted features are clustered to obtain the corresponding class labels to achieve sample pre-classification.Finally,the steganalysis network is trained separately using samples in each subclass to achieve more reliable steganalysis.We experimented on publicly available combined datasets of Bossbase1.01,Bows2,and ALASKA#2 with a quality factor of 75.The accuracy of our proposed pre-classification scheme can improve the detection accuracy by 4.84%for Joint Photographic Experts Group UNIversal WAvelet Relative Distortion(J-UNIWARD)at the payload of 0.4 bits per non-zero alternating current discrete cosine transform coefficient(bpnzAC).Furthermore,at the payload of 0.2 bpnzAC,the improvement effect is minimal but also reaches 1.39%.Compared with the previous steganalysis based on deep learning,this method considers the differences between the training contents.It selects the proper detector for the image to be detected.Experimental results show that the pre-classification scheme can effectively obtain image subclasses with certain similarities and better ensure the consistency of training and testing images.The above measures reduce the impact of sample content inconsistency on the steganalysis network and improve the accuracy of steganalysis.

A General Linguistic Steganalysis Framework Using Multi-Task Learning

Prevailing linguistic steganalysis approaches focus on learning sensitive features to distinguish a particular category of steganographic texts from non-steganographic texts,by performing binary classification.While it remains an unsolved problem and poses a significant threat to the security of cyberspace when various categories of non-steganographic or steganographic texts coexist.In this paper,we propose a general linguistic steganalysis framework named LS-MTL,which introduces the idea of multi-task learning to deal with the classification of various categories of steganographic and non-steganographic texts.LS-MTL captures sensitive linguistic features from multiple related linguistic steganalysis tasks and can concurrently handle diverse tasks with a constructed model.In the proposed framework,convolutional neural networks(CNNs)are utilized as private base models to extract sensitive features for each steganalysis task.Besides,a shared CNN is built to capture potential interaction information and share linguistic features among all tasks.Finally,LS-MTL incorporates the private and shared sensitive features to identify the detected text as steganographic or non-steganographic.Experimental results demonstrate that the proposed framework LS-MTL outperforms the baseline in the multi-category linguistic steganalysis task,while average Acc,Pre,and Rec are increased by 0.5%,1.4%,and 0.4%,respectively.More ablation experimental results show that LS-MTL with the shared module has robust generalization capability and achieves good detection performance even in the case of spare data.

A Deep Learning Driven Feature Based Steganalysis Approach

The goal of steganalysis is to detect whether the cover carries the secret information which is embedded by steganographic algorithms.The traditional ste-ganalysis detector is trained on the stego images created by a certain type of ste-ganographic algorithm,whose detection performance drops rapidly when it is applied to detect another type of steganographic algorithm.This phenomenon is called as steganographic algorithm mismatch in steganalysis.To resolve this pro-blem,we propose a deep learning driven feature-based approach.An advanced steganalysis neural network is used to extract steganographic features,different pairs of training images embedded with steganographic algorithms can obtain diverse features of each algorithm.Then a multi-classifier implemented as lightgbm is used to predict the matching algorithm.Experimental results on four types of JPEG steganographic algorithms prove that the proposed method can improve the detection accuracy in the scenario of steganographic algorithm mismatch.

基于元学习及域自适应的生成式文本隐写分析方法

生成式文本隐写方法能够生成流畅、自然的隐写文本,给文本隐写分析带来了巨大挑战。当源域(训练文本)与目标域(测试文本)使用相同隐写算法,且在相同语料库下训练时,基于神经网络的文本隐写分析方法具有较好的检测性能。然而在实际应用中,我们无法预测待检测文本是采用何种隐写算法及在何种语料库下训练所生成,导致大多数基于神经网络的文本隐写分析方法难以实用。为解决这个问题,引入了元学习及域自适应的基本思想来提高隐写分析模型的泛化检测能力。我们采用预训练语言表示模型RoBERTa构建了一个词重要性语义编码模块,以充分提取文本语义特征。针对所得特征,提出了一个词间关联多尺度感知模块来关注由隐写导致的存在于相邻词与非相邻词之间的词间关系变化。实验结果表明,在多跨域场景下,该方法相较于现有文本隐写分析方法Fs-Stega检测准确率平均提高了9%。

数字图像隐写分析综述

数字隐写是机密信息安全传递的重要方式,将机密信息隐藏于普通多媒体数据(图像或视音频)中可以实现隐蔽传输。而发现机密信息的隐蔽传输可采用隐写分析技术,隐写分析根据隐写引起的载体数据统计异常来判断多媒体数据是否含有秘密信息。近年来,隐写与隐写分析在相互对抗中不断进步与发展。随着社交网络的兴起,数字图像已成为社交媒介之一并广泛传播。本文以数字图像为例,梳理了近十余年数字图像隐写分析研究的发展现状;综述了传统隐写分析与深度学习隐写分析;探讨了各类方法面临的挑战,并展望了隐写分析的发展趋势。

Steganalysis Using Fractal Block Codes and AP Clustering in Grayscale Images

This paper presents a universal scheme （also called blind scheme） based on fractal compression and affinity propagation （AP） clustering to distinguish stego-images from cover grayscale images, which is a very challenging problem in steganalysis. Since fractal codes represent the ＂self-similarity＂ features of natural images, we adopt the statistical moment of fractal codes as the image features. We first build an image set to store the statistical features without hidden messages, of natural images with and and then apply the AP clustering technique to group this set. The experimental result shows that the proposed scheme performs better than Fridrich＇s traditional method.

Digital Image Steganographer Identification:A Comprehensive Survey

The rapid development of the internet and digital media has provided convenience while also posing a potential risk of steganography abuse.Identifying steganographer is essential in tracing secret information origins and preventing illicit covert communication online.Accurately discerning a steganographer from many normal users is challenging due to various factors,such as the complexity in obtaining the steganography algorithm,extracting highly separability features,and modeling the cover data.After extensive exploration,several methods have been proposed for steganographer identification.This paper presents a survey of existing studies.Firstly,we provide a concise introduction to the research background and outline the issue of steganographer identification.Secondly,we present fundamental concepts and techniques that establish a general framework for identifying steganographers.Within this framework,state-of-the-art methods are summarized from five key aspects:data acquisition,feature extraction,feature optimization,identification paradigm,and performance evaluation.Furthermore,theoretical and experimental analyses examine the advantages and limitations of these existing methods.Finally,the survey highlights outstanding issues in image steganographer identification that deserve further research.

An Improved Image Steganography Security and Capacity Using Ant Colony Algorithm Optimization

This advanced paper presents a new approach to improving image steganography using the Ant Colony Optimization(ACO)algorithm.Image steganography,a technique of embedding hidden information in digital photographs,should ideally achieve the dual purposes of maximum data hiding and maintenance of the integrity of the cover media so that it is least suspect.The contemporary methods of steganography are at best a compromise between these two.In this paper,we present our approach,entitled Ant Colony Optimization(ACO)-Least Significant Bit(LSB),which attempts to optimize the capacity in steganographic embedding.The approach makes use of a grayscale cover image to hide the confidential data with an additional bit pair per byte,both for integrity verification and the file checksumof the secret data.This approach encodes confidential information into four pairs of bits and embeds it within uncompressed grayscale images.The ACO algorithm uses adaptive exploration to select some pixels,maximizing the capacity of data embedding whileminimizing the degradation of visual quality.Pheromone evaporation is introduced through iterations to avoid stagnation in solution refinement.The levels of pheromone are modified to reinforce successful pixel choices.Experimental results obtained through the ACO-LSB method reveal that it clearly improves image steganography capabilities by providing an increase of up to 30%in the embedding capacity compared with traditional approaches;the average Peak Signal to Noise Ratio(PSNR)is 40.5 dB with a Structural Index Similarity(SSIM)of 0.98.The approach also demonstrates very high resistance to detection,cutting down the rate by 20%.Implemented in MATLAB R2023a,the model was tested against one thousand publicly available grayscale images,thus providing robust evidence of its effectiveness.

彩色图像隐写分析研究进展

传统的加密通信技术易被察觉,很难满足安全通信的需求。数字隐写术通过对载体修改来隐藏信息,可以实现隐蔽通信。但是,隐写术也可能被用于非法行为,使检测隐写术的隐写分析技术受到了越来越多的关注,因此隐写分析技术具有重大的研究意义。深度学习在计算机视觉、模式识别和自然语言处理等领域取得了很多研究成果,给隐写分析带来了新机遇,提出了新挑战,促使隐写分析产生新思想、新方法。彩色图像是互联网传输过程中的主流载体。但现有的彩色图像隐写分析特征主要依赖于人工设计,且隐写分析特征将彩色图像当成3个独立的灰度图像,并未充分考虑彩色图像三通道间的内部联系,对载密图像的检测能力有待提高。深度学习在彩色图像隐写分析领域中的应用也处于初步阶段。因此在介绍隐写和隐写分析的概念、分类和研究意义,概述其研究现状后,重点介绍了几种针对彩色图像的隐写分析关键技术,对其进行对比和总结,并分析其发展趋势。

面向隐写算法失配的小样本图像隐写分析方法

在实际的隐写分析应用场景中,待测隐写算法大多是未知的,难以获得足量带标记的样本,从而导致隐写算法失配问题.为提升在隐写算法未知且仅有少量标记图像时隐写分析的检测性能,提出新型隐写分析网络BTONet.首先,提出结合瓶颈注意力机制的改进SRNet,即BAMSRNet,作为BTONet的特征提取模块,从空间维度和通道维度对纹理区域进行关注,解决小样本环境下直接使用SRNet会导致检测性能不佳的问题,在带标记图像数量极少的情况下提取有辨识性的特征.然后,将正交投影损失和交叉熵损失有机结合,从特征和预测标签2个角度强化不同类别之间的正交性,提升分类模块的性能.最后,在隐写算法失配的情况下,将BTONet与4个经典空域深度隐写分析算法进行检测准确率、训练时长、测试时长和算法稳定性等方面的比较,并进行消融实验.实验结果表明:相较于目前先进的基于深度学习的隐写分析方法,BTONet在小样本环境下能够取得更优的检测性能,检测性能提升了1.02%~10.35%;同时取得了极佳的稳定性,将检测准确率方差降低至其他隐写算法的1/60~1/20.

基于深度可逆网络和差分编码的图像隐藏

伪装图像质量、恢复图像质量和传输安全性是图像隐藏最关注的3个问题。为解决这些问题,提出一种基于深度可逆网络和差分编码的图像隐藏方法,并用于保护电力巡检缺陷图像。首先训练深度可逆网络,利用训练好的可逆缩放网络对电力巡检缺陷图像进行向下缩放。与压缩感知等方法相比,可逆缩放网络能够恢复质量更高的缺陷图像。然后提出一种新的基于差分编码的嵌入算法,利用该算法将下缩放的缺陷图像嵌入到封面图像中。不同于现存很多方法直接对原图像像素值进行嵌入,所提方法先利用差分编码对缺陷图像进行编码,然后利用最低有效位算法完成嵌入操作,差分编码后的图像数值集中在更小的范围内,减少了嵌入对封面图像像素值的损害。实验结果表明,相较对比方法,所提方法伪装图像的峰值信噪比(PSNR)提高3.99 dB~16.56 dB,恢复缺陷图像的PSNR提高12.52 dB~17.02 dB。另外,该方法对SPAM的抗隐写分析性能优于对比方法。分析结果表明,所提方法在伪装图像质量、恢复缺陷图像质量和传输安全性方面的表现优于许多先进方法。

基于多尺度与注意力机制的图像隐写分析

针对目前隐写分析算法对图像复杂纹理区域特征表征能力较弱的问题,提出一种基于多尺度特征融合和注意力机制的隐写分析模型。该模型首先使用空域富模型滤波器对输入图像进行预处理,提取噪声成分残差,降低图像本身内容的影响;其次使用多尺度并行网络提取信号,增强对细微特征的学习;然后引入注意力机制对特征进行自适应加权,强调重要通道特征在分类中的作用,同时抑制非重要通道特征对分类的影响;最后提出一种协方差池化对深度神经网络学习后的各特征之间的相关性进行建模,并选取牛顿迭代法求解平方根矩阵,使网络训练更加高效。实验结果表明:在小波权重隐写算法0.5 bit/像素嵌入率的条件下,所提模型准确率达到了88.6%,证明了所提方法的有效性。

面向失配的图像隐写分析研究进展

尽管隐写分析在实验室环境下取得了显著的进步,但是在实际应用中,由于训练集和测试集的载体来源、隐写算法和嵌入率经常不同,导致隐写分析器性能下降,这种现象称为失配,严重阻碍了隐写分析的实际应用.因此,对目前面向失配问题的主要隐写分析方法进行了分析与总结.根据解决失配问题的思路,将现有失配隐写分析方法分为3类,即设计训练集、取证辅助和无监督领域适应,并对各类方法进行梳理和对比.基于对比结果,探讨了当前基于无监督领域适应的深度隐写分析模型面临的挑战以及未来的发展方向.研究结果表明:基于无监督领域适应的深度隐写分析模型是目前解决失配问题的最有效方案,领域对齐、中间域桥接、对抗学习等是设计该类深度隐写分析模型的主流思想;引入类别等细粒度信息以提高基于无监督领域适应的深度隐写分析模型的性能是未来研究的方向;针对不平衡样本及单/小样本等更恶劣的失配问题的解决方案仍待进一步探索.

全局协方差池化与多尺度特征融合的图像隐写检测

针对目前图像隐写深度检测模型中池化等操作造成特征图信息丢失,全局平均池化忽视高阶统计量的问题,提出一个基于全局协方差池化与多尺度特征融合的隐写检测模型:首先用多层小尺度卷积核替换多层感知器中的大尺度卷积核,增强特征表达能力的同时提高卷积计算效率;然后利用空洞卷积构建多尺度特征融合模块,减少模型在池化等过程中导致的细节特征信息损失;最后引入全局协方差池化,通过计算二阶统计量协方差作为最后的特征输出,增强检测模型对细节特征的捕捉能力。实验结果表明在不同的隐写算法和不同的嵌入率下,相比于近期典型的Xu-Net、Yedroudj-Net、Zhu-Net模型,所提模型的检测准确率均有显著提升,即使是与最新的Zhu-Net相比,准确率也提升了2.4%~7.3%。

基于尺寸变换的图像级特征增强隐写分析方法

随着深度学习的快速发展,基于深度学习的图像隐写分析技术研究取得了显著进展。然而,在残差特征提取及增强方面,传统图像预处理增强技术往往导致隐写信号的减弱,使得简单的图像预处理方法难以适配于隐写分析。对此,现有的深度学习隐写分析研究倾向于在不损害图像原有信息的基础上,设计固定的滤波核或对残差卷积层优化学习,缺乏对图像层面的隐写特征增强策略的可行性探讨。针对这一现象,提出了一种新颖高效的图像级特征增强隐写分析方法,通过最近邻插值算法扩大图像尺寸,在保留原始隐写信号的基础上进一步拓展分布相同的嵌入信号,从而增强模型对隐写残差特征的提取能力,无须对现有隐写分析流程做出显著改动即可有效提高隐写痕迹的可检测性。实验结果显示,所提方法能够显著提升模型在多种隐写算法下的检测准确率,尤其对于低嵌入率,其准确率最高可提升2.81%。该方法证实了图像层面预处理在隐写残差特征增强上的有效性,为深度学习隐写分析的图像残差特征提取提供了新的研究视角。

一种新的卷积神经网络图像隐写分析模型

针对现有卷积神经网络模型在图像隐写分析领域提取特征不充分、检测准确率不高的问题,提出一种融合转置卷积与普通卷积的图像隐写分析神经网络模型TCIS(Transposed Convolution-Convolutional Neural Network Image Steganalysis),包括四大模块:一是预处理模块,使用30个高通滤波器,从多个尺度提取图像噪声的残差信息,减少图像内容的影响;二是转置卷积模块,对特征图进行上采样,放大隐写特征;三是普通卷积模块,由卷积层、BN层和激活函数组成,卷积层包括5个,最后一层使用全局卷积的方式精简识别特征;四是分类模块,通过全连接层和Softmax层判断图像是否隐写。实验结果表明,相比于典型卷积神经网络图像隐写分析模型,TCIS模型在嵌入率0.4bpp情况下使用S-UNIWARD和HUGO算法的隐写分析准确率分别提升了2.94%~25.24%和3.92%~21.64%。

基于注意力机制的浅层图像隐写分析模型

为进一步提高隐写分析模型的检测准确率,提出一种基于注意力机制的浅层图像隐写分析模型,通过使用一个浅层神经网络控制模型参数量和训练时间,引入注意力模块,加速模型收敛,提升模型检测的准确率。实验结果表明,针对3种隐写算法在嵌入率分别为0.2 bpp和0.4 bpp时,检测准确率比浅层卷积神经网络(Shallow Convolution Neural Network,SCNN)均有提升,最高提升5.5%。

图像隐写的大数据分析方法

针对现有图像隐写分析方法存在的通用性较差、检测准确率欠佳等问题,提出了一种图像隐写的大数据分析方法。首先,根据自适应隐写算法的特性,截取待测图像的高复杂度区域作为核心分析图像。其次,利用数据库完成图像匹配,并得到匹配图像与核心分析图像间的隐写差异特征。最后,将特征输入卷积神经网络模型,完成隐写算法的检测与分类。实验结果表明,在图像全部匹配成功的条件下,该方法对6种隐写算法检测的平均准确率达到了93.98%,同时支持空域和频域的图像,具有较强的通用性。