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.     

基于块参照像素的无损信息隐藏算法

提出一种具有高嵌入容量的图像无损信息隐藏算法。首先将载体图像分成互不重叠的子块,然后在每块中选定一个参照像素,并计算参照像素与块内其它像素的差。在像素差直方图移位产生冗余空间之后,机密信息就可以无损地嵌入到这些冗余空间中。该方法在机密信息提取后可完全恢复载体图像,而且提取机密信息和恢复载体图像不需要除机密信息长度之外的任何信息。实验结果表明了该算法的有效性。     

A modified DWT-based image steganography technique

The main challenges of image steganography are imperceptibility of the cover image and no recoverability of the secret data. To deal with these challenges, a modified digital image steganography technique based on Discrete Wavelet Transform (DWT) is proposed. In proposed approach, two new concepts are being proposed to minimize the distortion in the cover image. The first one i.e. secret key computation concept is proposed to make it more robust and resistive towards steganalysis. The second one, known as blocking concept, is introduced to ensure least variation in the cover image. The proposed approach is tested over ten different cover images and two secret images. Its performance is compared with the six well-known steganography techniques. The experimental results reveal that the proposed approach performs better than the existing techniques in terms of imperceptibility, security and quality measures. The six image processing attacks are also applied on the stego-image to test the robustness of the proposed approach. The effects of compression, rotation, and application of different wavelets have also been investigated on the proposed approach. The results demonstrate the robustness of the proposed approach under different image processing attacks. Both stego-image and extracted secret images possess better visual quality.     

基于图像平滑度的空域LSB嵌入的检测算法

对图像像素值与邻域均值的差的分布进行建模,以该分布的方差定义了图像平滑度的概念,进而通过对消息嵌入、LSB平面取反带来的图像平滑度的变化进行分析,提出了针对空域LSB替换伪装算法的秘密消息长度估计算法,该算法可以准确地估计图像中嵌入的秘密信息数据量的大小.文中算法物理意义直观、实现简单、计算量小、计算速度快,有利于实现实时检测.实验结果表明该方法是有效的.     

An adaptive image steganographic scheme based on Noise Visibility Function and an optimal chaotic based encryption method

Steganography is the science of hiding secret message in an appropriate digital multimedia in such a way that the existence of the embedded message should be invisible to anyone apart from the sender or the intended recipient. This paper presents an irreversible scheme for hiding a secret image in the cover image that is able to improve both the visual quality and the security of the stego-image while still providing a large embedding capacity. This is achieved by a hybrid steganography scheme incorporates Noise Visibility Function (NVF) and an optimal chaotic based encryption scheme. In the embedding process, first to reduce the image distortion and to increase the embedding capacity, the payload of each region of the cover image is determined dynamically according to NVF. NVF analyzes the local image properties to identify the complex areas where more secret bits should be embedded. This ensures to maintain a high visual quality of the stego-image as well as a large embedding capacity. Second, the security of the secret image is brought about by an optimal chaotic based encryption scheme to transform the secret image into an encrypted image. Third, the optimal chaotic based encryption scheme is achieved by using a hybrid optimization of Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) which is allowing us to find an optimal secret key. The optimal secret key is able to encrypt the secret image so as the rate of changes after embedding process be decreased which results in increasing the quality of the stego-image. In the extracting process, the secret image can be extracted from the stego-image losslessly without referring to the original cover image. The experimental results confirm that the proposed scheme not only has the ability to achieve a good trade-off between the payload and the stego-image quality, but also can resist against the statistics and image processing attacks.     

Color local complexity estimation based steganographic (CLCES) method

In this paper we present the color local complexity estimation based steganographic (CLCES) method that is able of both preventing visual degradation and providing a large embedding capacity. A preprocessing stage is applied in the proposed scheme to improve the steganography security. The embedding capacity of each pixel is determined by the local complexity of the cover image, allowing good visual quality as well as embedding a large amount of secret messages. We classify the pixels using a threshold based on the standard deviation of the local complexity in the cover image to provide a compromise between the embedding capacity and the image visual quality. The experimental results demonstrated that the algorithm CLCES proposed produces insignificant visual distortion due to the hidden message. It provides a high embedding capacity that is superior respect to the offered by the existing schemes. The proposed method is a secure steganographic algorithm; it can resist the image quality measures (IQM) steganalysis attack. The RGB, YCbCr, and HSV color spaces are incorporated in the proposed scheme to ensure that the difference between the cover image and the stego-image which is indistinguishable by the human visual system (HVS). Finally, the proposed scheme is simple, efficient, and feasible for the adaptive steganographic applications.     

最优像素调整耦合基因算法的高容量图像隐写研究

为了解决当前图像隐写方案存在阶梯效应,使其不可感知能力差,且其信息隐藏容量小(≤50%)等不足,本文设计了最优像素调整耦合基因算法的高容量图像隐写术。基于HDWT(Haara Discrete Wavelet Transform) 机制,构造隐藏信息长度计算模型,找出图像分块的频域表示,以改善隐写鲁棒性;根据载体图像与隐写图像之间的绝对误差,设计适应度函数,借助基因算法,获取最优映射函数,将秘密信息嵌入到HDWT系数中;并设计最优像素变换方案,降低载秘图像与载体图像之间的嵌入误差,显著增大隐写容量;再设计该算法的提取机制,获取信息图像;以PSNR(Peak Signal to Noise Ratio)构建反馈机制,优化提取质量。仿真结果显示：与其他隐写机制相比,本文算法具备更大的隐写容量和更强的不可感知性能;拥有更高的检测精度,可有效区分载体与隐写图像特征值。     

基于图像位平面相关性的隐写分析*

通过对原始图像和经过LSB隐写的隐蔽图像的分析,信息的再次嵌入对隐蔽图像与其位平面之间相关性的影响很小,而对原始图像与其位平面之间相关性的影响很大。以此为基础,提出了一种基于图像和其位平面之间的线性相关性的隐写分析方法,实现了对图像是否载密的检测。实验表明,对于图像的LSB隐写,可以进行有效的检测。该方法也可以对图像的其他位平面的嵌入进行分析。     

Using contourlet transform and cover selection for secure steganography

In this paper, we present a new adaptive contourlet-based steganography method that hides secret data in a specific or automatically selected cover image. Our proposed steganography method primarily decomposes the cover image by contourlet transform. Then, every bit of secret data is embedded by increasing or decreasing the value of one coefficient in a block of a contourlet subband. Contourlet coefficients are manipulated relative to their magnitudes to hide the secret data adaptively. In addition to proposing contourlet-based steganography method, this work investigates the effect of cover selection on steganography embedding and steganalysis results. We demonstrate, through the experiments, that choosing suitable cover image by a proper selection measure could help the steganographer reduce detectability of stego images. The proposed technique is examined with some state-of-the-art steganalysis methods, and the results illustrate that an image can successfully hide secret data with average embedding capacity of 0.02 bits per pixel in a random selected cover image. Cover selection improves the embedding capacity up to 0.06 bits per pixel. Several experiments and comparative studies are performed to show the effectiveness of the proposed technique in enhancing the security of stego images, as well as to demonstrate its gain over the previous approaches in literature.     

利用模运算及其周期性特点的安全隐写算法

为了提供较大的可调信息嵌入量和保持载密图像良好的视觉质量,提出一种基于模运算及其周期性特点的安全隐写算法。首先将秘密信息流转为n进制信息流表示,然后根据余数循环的特点,应用具有安全特性的模函数将一位n进制信息嵌入载体图像的一个像素中,直至信息嵌入完毕。选择不同的参数n可以得到不同的嵌入率和载密图像视觉质量。理论分析和实验结果表明,本文算法在保持高嵌入率的同时仍保持较好的载密图像视觉质量,且安全性好;其信息嵌入率选择具有高度弹性,可以根据实际要求选择在嵌入率和载密图像质量都很理想的隐写方案。与同类算法相比具有更强的实用性。     

Inverted pattern approach to improve image quality of information hiding by LSB substitution

Capacity and invisibility are two targets of the methods for information hiding. Because these two targets contradict each other, to hide large messages into the cover image and remain invisible is an interesting challenge. The simple least-significant-bit (LSB) substitution approach, which embeds secret messages into the LSB of pixels in cover images, usually embeds huge secret messages. After a large message is embedded, the quality of the stego-image will be significantly degraded. In this paper, a new LSB-based method, called the inverted pattern (IP) LSB substitution approach, is proposed to improve the quality of the stego-image. Each section of secret images is determined to be inverted or not inverted before it is embedded. The decisions are recorded by an IP for the purpose of extracting data and the pattern can be seen as a secret key or an extra data to be re-embedded. The experimental results show that our proposed method runs fast and has better results than that of previous works.     

基于可容忍失真范围的数字图像隐写方法研究

以往的基于可容忍失真范围的隐写方案可以将隐写引起的失真控制在一定范围,但需要同时具备含密图像与原始图像才能提取秘密信息。提出了一种新的基于可容忍失真范围的隐写方案,该方案以一定质量因子的JPEG压缩作为可容忍失真,嵌入信息时仅在可容忍失真范围内改动原始图像,并具有含密图像的JPEG压缩版本与原始图像的JPEG压缩版本完全一致的特性,因此接收端不需原始图像,只要将含密图像与其JPEG压缩版本相减,便可提取出秘密信息,克服了原有此类方法需要原始图像才能提取秘密信息的缺点。实验结果表明,用该方案隐写所得到含密图像不仅在质量上优于其对应的JPEG压缩版本的质量,而且具有一定的抗分析安全性。     

新的基于小波对比度的LSB密写

利用图像块的小波对比度可以判定图像各像素的LSB（Least Significant Bit）嵌入深度,以提高嵌入容量并保持载密图像的质量。提出了一种新的基于小波对比度的密写方案,将图像块像素的低4位置0后进行小波变换并计算小波对比度作为嵌入深度的判决信息,由嵌入深度逐块逐像素密写。实验结果表明,该方法不仅保持了较高的嵌入容量,而且提取秘密信息时不需任何附加信息。     

小波域最小嵌入失真函数设计及其在隐写中的应用

为了提高隐写方案的安全性,提出一种基于最小嵌入失真原理和网格码的图像隐写算法。首先在离散小波域结合人眼视觉特性和整数提升小波变换设计了失真测度函数,主要考虑了亮度、频率和纹理掩蔽因子对载体失真的影响。然后结合网格码设计了隐写算法,将嵌入信息对载体的修改最小化并且集中在人眼不敏感区域。实验结果显示,方案具有良好的视觉不可见性,且能抵抗空域、小波域等隐写分析的攻击,安全容量达到0.4 bits/pixel。     

Secret image sharing with steganography and authentication

A novel approach to secret image sharing based on a (k,n)-threshold scheme with the additional capabilities of steganography and authentication is proposed. A secret image is first processed into n shares which are then hidden in n user-selected camouflage images. It is suggested to select these camouflage images to contain well-known contents, like famous character images, well-known scene pictures, etc., to increase the steganographic effect for the security protection purpose. Furthermore, an image watermarking technique is employed to embed fragile watermark signals into the camouflage images by the use of parity-bit checking, thus providing the capability of authenticating the fidelity of each processed camouflage image, called a stego-image. During the secret image recovery process, each stego-image brought by a participant is first verified for its fidelity by checking the consistency of the parity conditions found in the image pixels. This helps to prevent the participant from incidental or intentional provision of a false or tampered stego-image. The recovery process is stopped if any abnormal stego-image is found. Otherwise, the secret image is recovered from k or more authenticated stego-images. Some effective techniques for handling large images as well as for enhancing security protection are employed, including pixelwise processing of the secret image in secret sharing, use of parts of camouflage images as share components, adoption of prime-number modular arithmetic, truncation of large image pixel values, randomization of parity check policies, etc. Consequently, the proposed scheme as a whole offers a high secure and effective mechanism for secret image sharing that is not found in existing secret image sharing methods. Good experimental results proving the feasibility of the proposed approach are also included.     

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。     

High payload image steganography scheme with minimum distortion based on distinction grade value method

Presently, the design and development of an effective image steganography system are facing several challenges including the low capacity, poor robustness and imperceptibility. To surmount such limitations, it is important to improve the capacity and security of the steganography system while maintaining a high signal-to-noise ratio (PSNR). Based on these factors, this study is aimed to design and develop a distinction grade value (DGV) method to effectively embed the secret data into a cover image for achieving a robust steganography scheme. The design and implementation of the proposed scheme involved three phases. First, a new encryption method called the shuffle the segments of secret message (SSSM) was incorporated with an enhanced Huffman compression algorithm to improve the text security and payload capacity of the scheme. Second, the Fibonacci-based image transformation decomposition method was used to extend the pixel’s bit from 8 to 12 for improving the robustness of the scheme. Third, an improved embedding method was utilized by integrating a random block/pixel selection with the DGV and implicit secret key generation for enhancing the imperceptibility of the scheme. The performance of the proposed scheme is assessed experimentally to determine the imperceptibility, security, robustness and capacity. The resistance of the proposed scheme is tested against the statistical, χ², Histogram and non-structural steganalysis detection attacks. The obtained PSNR values revealed the accomplishment of the higher imperceptibility and security by the proposed DGV scheme while maintaining higher capacity compared to the reported findings. In short, the proposed steganography scheme outperformed the commercially available data hiding schemes, thereby resolved the existing issues.

     

使用四叉树分割进行自适应空域隐写

目的 针对自适应隐写术可有效避免对载体敏感区大量修改的关键问题,为间接提高安全性和增大隐写容量,在四叉树分割和自适应像素对匹配（APPM）的基础上提出一种自适应空域隐写术。方法 首先该方法以图像块的纹理复杂度作为一致性测度并且设置图像块大小为判别准则进行图像分割,根据四叉树分割结果中面积较小的图像块属于复杂区域,较大的属于平滑区域,按照图像块面积大小将图像分成由高复杂、中复杂、低复杂三大区域构成。其次嵌密方式采用APPM,根据密信容量和载体图像选择进制数B。最后,为了保证安全性和提高容量,优先选择高复杂区嵌入不低于B进制的密信,在中复杂区进行B进制的密信嵌入,在低复杂区选择不高于B进制的密信嵌入。结果 为了验证提出的方法,选8幅经典图作为实验,在嵌入率1.92 bit/pixel的情况下,与已有PVD系列算法和DE算法相比具有更高的PSNR值,PSNR值高达48 dB。此外与APPM算法比较,在嵌入率2.5 bit/pixel情况下,该算法的平均KL距离相比传统APPM算法减小了25.37%,平均一阶Markov安全指标值相比传统APPM算法减小了12.11%,对应的平均PSNR值相比传统APPM算法提高0.43%,在嵌入率1.5 bit/pixel情况下,该算法的平均KL距离相比传统APPM算法减小了37.84%,平均一阶Markov安全指标值相比传统APPM算法减小了26.61%,对应的平均PSNR值相比传统APPM算法提高1.56%。此外,从RSP图库中随机选1 000幅图作为数据集,在嵌入率0.5,0.6,0.7,0.8,0.9和1.0 bit/pixel条件下,结合SPAM特征和SVM分类器的最小平均错误率均高于LSB系列经典算法和APPM算法。结论 1）考虑了人类视觉系统对图像不同区域的敏感性不同,通过对图像进行四叉树分割预处理,优先选择非敏感区进行隐写,保证了一定的安全性要求,低嵌入率下抗SPAM检测和统计不可见性方面比较有优势。2）在四叉树分割中,对于隐写前后图像的四叉树分割结果不同的异常情况,采用一种图像块纹理复杂度调整方案,保证了密信正确完整提取。3）利用了APPM算法的大容量特性,可以隐写嵌入率大于1 bit/pixel的密信,比较适用于大容量的密信隐写,而且可以嵌入任意进制的密信,最大程度地减少嵌入失真,此外,进行了四叉树分割预处理,在安全性方面优于传统APPM算法。     

面向可逆图像处理网络的可证安全自然隐写

目的 自然隐写是一种基于载体源转换的图像隐写方法,基本思想是使隐写后的图像具有另一种载体的特征,从而增强隐写安全性。但现有的自然隐写方法局限于对图像ISO(International Standardization Organization)感光度进行载体源转换,不仅复杂度高,而且无法达到可证安全性。为了提高安全性,本文结合基于标准化流的可逆图像处理模型,在隐空间完成载体源转换,同时通过消息映射的设计做到了可证安全的自然隐写。方法 利用目前发展迅速的基于可逆网络的图像处理方法将图像可逆地映射到隐空间,通过替换使用的隐变量完成载体源的转换,从而避免对原始图像复杂的建模。同时,改进了基于拒绝采样的消息映射方法,简单地从均匀分布中采样以获得需要的条件分布,高效地将消息嵌入到隐变量中,并且保证了嵌入消息后的分布与原本使用的分布一致,从而实现了可证安全的自然隐写。结果 针对图像质量、隐写容量、消息提取准确率、隐写安全性和运行时间进行了实验验证,结果表明在使用可逆缩放网络和可逆去噪网络时能够在每个像素值上平均嵌入5.625 bit消息,且具有接近99%的提取准确率,同时隐写分析网络SRNet(st...     

引入超分辨率下采样误差的图像边信息估计隐写

目的 由于空域图像下采样过程中提供的量化误差边信息能够有效提升隐写安全性,为了得到下采样之前的高分辨率图像,提出一种基于超分辨率网络的空域图像边信息估计隐写方法。方法 受原始下采样边信息隐写方法的启发,使用超分辨率网络生成被称为预载体的高分辨率图像。同时利用现有的空域图像对称失真算法得到每个像素点的修改失真,然后以浮点型精度对预载体下采样,得到和载体同分辨率的图像形式,利用对应像素点间的差值指导像素点的修改方向,实现基于初始失真的非对称失真调整。首先以峰值信噪比和极性估计准确率为指标对比了多种超分辨率网络以及基于传统插值方法的上采样性能,并通过调整初始失真分别进行隐写和隐写分析实验,选择使安全性提升最大的残差通道注意力机制网络及其对应调整系数作为本文的下采样边信息估计隐写方法。结果 使用隐写领域中常用的3个数据库、两种传统初始失真函数以及两类隐写分析方法进行实验。在跨数据集的隐写安全性上,相比传统隐写方法,在对抗基于手工特征和基于深度学习的隐写分析时,本文方法的安全性均有显著提升,如在测试集载体图像上,嵌入率为0.5 bit/像素时,安全性分别提升6.67%和6.9%;在训练集载体图像上,本文方法的安全性在比传统方法有很大提升的基础上,甚至在一些情况下能够高于原始边信息隐写方法的安全性,如在对抗基于手工特征的隐写分析器且嵌入率为0.1 bit/像素时,安全性提升1.08%;在对抗基于深度学习的隐写分析器且嵌入率为0.5 bit/像素时,安全性提升0.6%。结论 实验表明,使用超分辨率网络作为下采样边信息估计的工具,并利用估计边信息调整嵌入修改的初始失真,能够有效提升传统隐写方法的安全性,并接近甚至在部分情况下超越了原始边信息隐写的安全性。除此之外,本文方法与原始边信息隐写方法具有不同的修改模式,而且具有更广泛的适用性。