Recently, coverless steganography algorithms have attracted increased research attention due to their ability to completely resist steganalysis algorithms. However, the existing algorithms do not attain the same robust balance against geometric and non-geometric attacks. In addition, most of the existing methods need to transmit some auxiliary information along with the stego-images, which increases the cost of the hidden information. In this paper, a robust coverless image steganography algorithm based on a hash generation model is proposed. Different from the existing methods, the hash sequences are generated by an end-to-end CNN model, where the input is the original images, and the output is the corresponding hash sequences. Therefore, no auxiliary information needs to be transmitted when hiding the secret information. Moreover, the attention mechanism and adversarial training are introduced to improve the robustness of the model. The loss function is redesigned to accommodate these operations. Finally, an index structure is built to enhance the mapping efficiency. The experimental results show that the proposed method possesses better robustness and security compared with the state-of-the-art coverless image steganography algorithms.

High Efficiency Video Coding (HEVC) places great emphasis on optimizing compression efficiency, where compression efficiency denotes file size ratio before and after compression. The current HEVC steganography is prone to cause degradation in compression efficiency. To analyze and avoid this problem, a Steganographic Compression Efficiency Degradation Model (SCEDM) is first proposed, which leverages the area ratio of different types of Coding Units (CU) as the distribution of block partitioning structure and combines with the K-L divergence to describe the compression efficiency degradation. By minimizing the output of the SCEDM, the degradation of compression efficiency caused by steganographies can be minimized. Besides, it is also proved that this minimizing process will not increase extra visual quality distortion. Based on this model, a novel adaptive steganography using HEVC intra block partitioning structure is proposed. This steganography consists of three parts: the CU Depth based Hierarchical Coding (CDHC) method, the structure merging strategy and the adaptive matching method. The CDHC method can convert secret binary bits to different block structures. The structure merging strategy improves the capacity, and the adaptive matching method minimizes the compression efficiency degradation according to the proposed SCEDM. The proposed steganography is further compared with state-of-the-art steganographies to confirm the effectiveness and advantages of the proposed model and steganography in compression efficiency, capacity, visual quality and resistance to video steganalysis.

Calibration is a common method for steganalysis, and Intra Prediction Mode (IPM) shift is a typical phenomenon used in calibration to detect video steganography. The current HEVC steganography lacks resistance to steganalysis based on this phenomenon because the new technology of HEVC introduces steganographic distortion in addition to providing more potential steganographic space. In this paper, an HEVC steganographic algorithm that resists IPM shift is proposed. First, we introduce the IPM shift in HEVC, and the previous H.264 steganalytic IPM shift feature is modeled and improved. By analyzing the HEVC encoding process, we found that modifying large-size blocks has a more significant impact on compression efficiency, while small ones are more sensitive to IPM optimality. Therefore, we perform the embedding channel division based on the block size and design the distortion function separately. In addition, we discover a unique IPM transition probability distribution in HEVC. According to our analysis, this unique distribution arises due to HEVC's MPM rules and the regularity of IPM direction. Modifying IPM in HEVC will change such distribution, thus, a mapping rule is designed based on this distribution to achieve a better embedding effect. Experimental results show that the channel division and proposed distortion function can effectively improve the overall performance. The proposed steganography outperforms the state-of-the-art steganography in resisting steganalysis, bitrate controlling, and visual quality.

High efficiency video coding (HEVC) is the latest high-performance video coding standard, and HEVC video steganography has become a new way to hide data for covert communication. This paper proposes a novel multilevel steganography algorithm based on diamond-encoded prediction unit (PU) partition modes. The PU modes of smaller$8\times 8$and$16\times 16$CUs are selected as carriers for information hiding. The diamond-coding rules are adopted to enhance the expressive ability of limited PU types, allowing them to carry more information under limited modification. Based on the encoded PU partition modes, three different embedding levels with different capacities are proposed. This paper’s most outstanding contribution is the introduction of convolutional neural networks (CNNs) for the first time to improve visual quality and reduce steganographic video bitrate increases. Experimental results show that the embedding capacity of the proposed algorithm is significantly higher than the state-of-the-art work at the same bitrate, whether in high- or low-resolution HEVC videos. At the same time, the visual quality of the steganographic videos is excellent, and the resistance to video steganalysis is strong.

Nowadays, the steganography of HEVC(High Efficiency Video Coding) videos has been widely researched. The existing steganography algorithms mainly focused on eliminating inter-block distortion and only embedded data in small-sized blocks, which did not take intra-block distortion into consideration and wasted large embedding capacity. In this paper, a HEVC steganography algorithm based on DCT/DST coefficients is proposed. It uses the intra-frame error propagation-free method to embed data in all blocks for eliminating inter-block distortion and especially, introduces CNN(Convolutional Neural Network) to reduce intra-block distortion. Experimental results show that the proposed algorithm improves the embedding capacity greatly and performs well on visual quality, bitrate increasing and anti-steganalysis.

The classical linear frequency modulated waveform (LFMW) for synthetic aperture radar (SAR) can achieve a high range resolution and a superior signal-to-noise ratio (SNR) for point targets. Therefore, numerous researches got increasingly passionate about increasing resolution and SNR, rather than information acquisition of LFMW. This paper quantitatively analyzes and evaluates target information acquisition ability of LFMW, which is conducive to target perception. Firstly, a factor contributing to information acquisition of LFMW, is obtained under the discretized signal model. Secondly, a connection between LFMW and eigenvectors of target feature matrix is derived through the discrete Fourier bases, thus the information acquisition of LFWM is presented. Finally, the numerical simulations are made for the stochastic extended targets information acquisition of LFMW, and the results are employed to verify the theory analysis.

Abstract Inter‐frame forgery is a common type of video forgery in digital videos. In this paper, a method based on the consistency of quotient of mean structural similarity (QoMSSIM) is proposed. For original videos, the QoMSSIM are consistent, but in forgeries the consistency will be destroyed. First, the mean structural similarity (MSSIM) between every two adjacent frames is extracted, and then the quotients between every two sequential MSSIM are calculated. Finally, the quotient of mean SSIM after post‐processing, normalization and quantization is used as distinguishing feature to identify inter‐frame forgeries. Experiments are conducted on a large database and support vector machine (SVM) is used to distinguish original videos and inter‐frame forgeries. Experimental results show that the proposed method is efficient in differentiating original videos and forgeries. For differentiating frame deletion and insertion forgeries, the proposed method performs also pretty well. Compared with the other method, the proposed method has higher classification accuracy, lower computational complexity and robustness against recompression and white Gaussian noise. Copyright © 2016 John Wiley & Sons, Ltd.

ABSTRACT Frame insertion and deletion are common inter‐frame forgery in digital videos. In this paper, an efficient method based on quotients of correlation coefficients between local binary patterns (LBPs) coded frames is proposed. This method is composed of two parts: feature extraction and abnormal point detection. In the feature extraction, each frame of a video is coded by LBP. Then, quotients of correlation coefficients among sequential LBP‐coded frames are calculated. In the abnormal point detection, insertion and deletion localization is achieved by using Tchebyshev inequality twice followed by abnormal points detection based on decision‐thresholding. Experimental results show that our method has high detection accuracy and low computational complexity. Copyright © 2014 John Wiley & Sons, Ltd.

ABSTRACT With the ongoing development of rendering technology, computer graphics (CG) are sometimes so photorealistic that to distinguish them from photographic (PG) images by human eyes has become difficult. To this end, many methods have been developed for automatic CG and PG classification. In this paper, we present a simple, yet efficient, multiresolution approach to distinguish CG from PG based on uniform gray‐scale invariant local binary patterns (LBPs) with the help of support vector machines (SVM). We select YCbCr as the color model. The original Joint Photographic Experts Group (JPEG) coefficients of Y, Cb, and Cr components and their prediction errors are used for two LBP operators. From each 2D array and each LBP operator, we obtain 59 uniform LBP features. In total, 12 groups of 59 features are obtained from each image. But after multiresolution analysis, we select six groups of 59 features for CG and PG classification. The proposed features have been tested with thousands of CG and PG. Classification accuracy reaches 95.1% with support vector machines and outperforms the state‐of‐the‐art works. Copyright © 2013 John Wiley & Sons, Ltd.