Steganalysis of images based on spatial domain and two-dimensional JPEG array

The aim of digital image steganalysis is to detect hidden information (which can be a message or an image) in a steganographic image. An ideal steganography method encrypts the information in the image such that it cannot be easily detected. Currently, a wide variety of different steganography techniques are being used; therefore, more advanced steganalysis methods are needed that can detect the steganographic images coded by different techniques. A typical steganalysis technique consists of two parts: (1) feature extraction and (2) classification. In this paper, a new steganalysis technique based on the Markov chain process is proposed. In the proposed technique, after extraction of the new features, a non-linear classifier named support vector machine is applied to classify clean and encrypted images. Analysis of variance is used to reduce the dimensions of the proposed features. The performance of the proposed technique is compared against subtractive DCT coefficient adjacency matrix (SDAM) and subtractive pixel adjacency matrix (SPAM) methods using an image database prepared by three strong steganography techniques called yet another steganographic scheme, model based, and perturbed quantization. The obtained results show that the proposed method provides better performance than SDAM and SPAM methods.     

Steganalysis by subtractive pixel adjacency matrix and dimensionality reduction

Subtractive pixel adjacency matrix(SPAM)features,introduced by Pevn′y et al.as a type of Markov chain features,are widely used for blind steganalysis in the spatial domain.In this paper,we present three improvements to SPAM as follows:1)new features based on parallel subtractive pixels are added to the SPAM features,which only refer to collinear subtractive pixels;2)features are extracted not only from the spatial image,but also from its grayscale-inverted image,making the feature matrices symmetrical and reducing their dimensionality by about half;and 3)a new kind of adjacency matrix is used,thereby reducing about 3/4 of the dimensionality of the features.Experimental results show that these methods for dimensionality reduction are very effective and that the proposed features outperform SPAM.     

基于MED信道选择和虚拟嵌入块的YASS改进算法

为抵抗YASS隐写分析算法,本文提出一种新的YASS改进算法。该算法首先利用密钥选择不规则的区域,生成一个虚拟的8×8嵌入块;然后根据图像自身特点,提出一种最小化嵌入失真的计算方法,对嵌入块修改后的量化DCT系数进行失真分析,选择失真影响最小的交流DCT系数进行秘密信息嵌入,取代传统方法直观选择前19交流DCT系数。将文中改进算法与虚拟嵌入块YASS（VH-YASS）算法进行了对比实验,实验结果证明,依据嵌入失真由小到大的顺序选择嵌入信道,在抵抗隐写分析和视觉质量方面都优于VH-YASS隐写方法。     

几种基于JPEG图像的隐写术比较研究

随着计算机网络的快速发展和应用,信息安全问题日益突出。在这样的情况下,出现了一种隐藏通信技术——隐写术,深入研究Outguess、F5和YASS算法,通过大量样本分析,掌握其算法的本质和加密图像的特点。通过对这几种算法的深入研究和对比,简要对隐写术的发展和应用做了探讨。     

基于改进局部二值模式的YASS检测方法

如今互联网已经逐渐渗透到人们生活的诸多方面,成为日常通信的重要途径。信息隐藏作为一种通信技术,通过将秘密信息嵌入常见载体中以达到隐蔽通信的目的。图像隐写是利用图像作为载体进行信息隐藏的一门技术与科学,YASS（Yet Another SteganographicScheme that Resists Blind Steganalysis）通过随机选取图像的子块进行DCT变换和QIM信息嵌入,具有较高的安全性。文中通过引入图像的局部二值模式（LBP）这一概念,根据YASS算法特点,分析图像的局部纹理变化,改进局部二值模式,利用局部有序对比模式（LOCP）的特征进行隐写分析。通过大量实验表明,相比传统的YASS隐写分析,文中所提方法在分析检测正确率等方面都有更好的效果。     

A channel selection rule for YASS

In this paper, a channel selection rule for YASS （Yet-Another-Secure-Steganography） is proposed. Secret message embedding imposes distortion to the cover image. The larger the distortion, the less secure the steganographic algorithm. Our channel selection rule engages in minimizing this distortion brought in by YASS. In our rule, the distortion caused by unit change on each quantized DCT （Discrete Cosine Transformation） component is computed, and the components with smaller unit change distortion are selected with higher priority. This channel selection rule reduces distortion to the medium spatial domain image and the final JPEO image. Experimental results show that our improved YASS scheme outperforms original YASS scheme on the aspects of both perception and statistics. This new channel selection rule can also be combined with other enhancements in YASS framework to further boost the performance.