Search

Due to their good rotational invariance and stability, image continuous orthogonal moments are intensively applied in rotationally invariant recognition and image processing. However, most moments produce numerical instability, which impacts the image reconstruction and recognition performance. In this paper, a new set of invariant continuous orthogonal moments, polar harmonic Fourier moments (PHFMs), free of numerical instability is designed. The radial basis functions (RBFs) of the PHFMs are much simpler than those of the Chebyshev-Fourier moments (CHFMs), orthogonal Fourier-Mellin moments (OFMMs), Zernike moments (ZMs), and pseudo-Zernike moments (PZMs). For the same degree, the RBFs of the PHFMs have more zeros and are more evenly distributed than those of the ZMs and PZMs. Therefore, PHFMs do not suffer from information suppression problem; hence, the image description ability of the PHFMs is superior to that of the ZMs and PZMs. Moreover, the RBFs of the PHFMs are always less than or equal to 1.0 near the unit disk center, whereas those of the OFMMs, PZMs, CHFMs, and radial harmonic Fourier moments (RHFMs) are infinite (implying numerical instability). This indicates that PHFMs can outperform these moments in image reconstruction tasks. We theoretically and experimentally demonstrate that PHFMs outperform the above moments in reconstructing images and recognizing rotationally invariant objects considering noise and various attacks. This paper also details the significance of the PHFM phase in image reconstruction, angle estimation using PHFMs, and the accurate moment selection of the PHFMs.

In recent years, with the improvements in machine learning, image forensics has made considerable progress in detecting editing manipulations. This progress also raises more questions in image forensics research, such as can the parameters applied in a manipulation be estimated. Many parameter estimation works have already been performed. However, most of these works are based on mathematical analyses. In this paper, we attempt to solve a particular parameter estimation problem from a different aspect. Specifically, a new convolutional neural network (CNN) model is proposed to estimate the resampling rate for resampled images regardless of whether the image is upscaled or downscaled. This model features an original layer to generate a measurable energy map toward the estimation of resampling rate (METEOR). The METEOR layer is demonstrated to be an outstanding method that can assist in enhancing the estimation performance of the CNN. Furthermore, the METEOR layer can also increase the robustness of the CNN against JPEG compression, which makes it extremely important in realistic application scenarios. Our work has verified that machine learning, particularly CNNs, with proper optimization can also be refined to adapt to parameter estimation in digital forensics with excellent performance and robustness.

Utilizing statistical models of binary images is a common and effective means to steganalyze binary images, and the design of the statistical model is essential to the performance of steganalysis. In this paper, we propose a new model based on a histogram of pixel structuring elements (SEs), which is a suitable representation of a binary image for the task of steganalysis. The texture property and the dependency among pixels are considered inside the SEs. The SEs with different patterns will be evaluated comprehensively according to a statistical criterion, and some of them will be selected to construct the feature set for training the steganalyzer. The distributions of these selected SEs, which contain many highly flippable pixels, will be emphasized by the criterion, and they can reflect the difference between cover images and stego-images. Finally, a series of experiments are conducted on two datasets, and the results show that the proposed scheme significantly outperforms state-of-the-art schemes.

Reversible data hiding (RDH) has unique advantage in copyright and integrity protection for multimedia contents. As a typical RDH scheme, histogram shifting technique (HS) has found wide applications due to its high quality of marked image. At present, most existing HS-based RDH schemes rely on single histogram generated from cover image to hide data. Since the single histogram-based approach (SH_RDH) commonly employs smooth regions in the cover image for data hiding, it might not well utilize the cover image and exploit the correlations among image contents of different texture characteristics. In this paper, a novel RDH general framework using multiple histograms modification (MH_RDH) is proposed, which involves two key issues as follows: 1) the construction of multiple histograms based on optimized multi-features and 2) the rate allocation among multiple histograms is formulated as the one of rate-distortion optimization and solved with evolutionary algorithms. The experimental results show that the proposed method could considerably increase the payload of current MH_RDH-based embedding (ranging from 0.2 to 0.7 bpp for most test images) and outperform the other state-of-the-art SH_RDH and MH_RDH schemes.

Detection of double Joint Photographic Experts Group (JPEG) compression is an important part of image forensics. Although methods in the past studies have been presented for detecting the double JPEG compression with a different quantization matrix, the detection of double JPEG compression with the same quantization matrix is still a challenging problem. In this paper, an effective method to detect the recompression in the color images by using the conversion error, rounding error, and truncation error on the pixel in the spherical coordinate system is proposed. The randomness of truncation errors, rounding errors, and quantization errors result in random conversion errors. The pixel number of the conversion error is used to extract six-dimensional features. Truncation error and rounding error on the pixel in its three channels are mapped to the spherical coordinate system based on the relation of a color image to the pixel values in the three channels. The former is converted into amplitude and angles to extract 30-dimensional features and 8-dimensional auxiliary features are extracted from the number of special points and special blocks. As a result, a total of 44-dimensional features have been used in the classification by using the support vector machine (SVM) method. Thereafter, the support vector machine recursive feature elimination (SVMRFE) method is used to improve the classification accuracy. The experimental results show that the performance of the proposed method is better than the existing methods.

In this paper, a binary image steganographic scheme is presented, which aims at minimizing the embedding distortions measured by prediction. A prediction model of the center pixel’s value is established in a $3 \times 3$ local region. A concept of “uncertainty” is introduced to represent the prediction result and the uncertainty is defined as the proximity of probabilities about whether the center pixel is black or white. A pixel with high uncertainty means that it is hard to distinguish whether it has been flipped or not, and thus the distortion introduced by flipping this pixel is small. The uncertainty is an appended statistical explanation of human visual perception and the distortion measurement based on it can evaluate the embedding changes on both vision and statistics. Benefiting from the statistics, uncertainty can evaluate the distortion influence in an extended local region. To play the advantage of distortion measurement, the syndrome-trellis code (STC) is employed to minimize the embedding distortions. Comparisons with prior schemes demonstrate that the proposed steganographic scheme achieves high vision imperceptibility and statistical security.

In recent years, fingerprint authentication systems have been extensively deployed in various applications, including attendance systems, authentications on smartphones, mobile payment authorizations, as well as various safety certifications. However, similar to the other biometric identification technologies, fingerprint recognition is vulnerable to artificial replicas made from cheap materials, such as silicon, gelatin, etc. Thus, it is especially necessary to distinguish whether a given fingerprint is a live or a spoof one prior to such authentication. In order to solve the problems above, a novel local descriptor named Weber local binary descriptor for fingerprint liveness detection (FLD) has been proposed in this paper. The method consists of two components: the local binary differential excitation component that extracts intensity-variance features and the local binary gradient orientation component that extracts orientation features. The co-occurrence probability of the two components is calculated to construct a discriminative feature vector, which is fed into support vector machine (SVM) classifiers. The effectiveness of the proposed method is intuitively analyzed on the image samples and numerically demonstrated by Mahalanobis distance. Experiments are performed on two public databases from FLD competitions from 2011 and 2013. The results have proved that the proposed method obtains the best detection accuracy among the existing image local descriptors in FLD.

Resampling detection is one of the most important topics in image forensics, and the most widely used method in resampling detection is spectral analysis. Since JPEG is the most widely used image format, it is reasonable that the resampling operation is processed on JPEG images. JPEG block artifacts bring severe interference to spectrum-based methods and degrade the detection performance. In addition, the spectral characteristics of the downscaling scenarios are very weak. The detection of downscaling still presents a considerable challenge to forensic applications. In this paper, we propose a method to estimate the downscaling factors of pre-JPEG compressed images in the presence of image downscaling after JPEG compressions. We first analyze the spectrum of scaled images and give an exact formulation of how the scaling factors influence the appearance of periodic artifacts. The expected positions of the characteristic resampling peaks are analytically derived. For the downscaling scenario, the shifted JPEG block artifacts produce periodic peaks, which cause misdetection in the characteristic peak. We find that the interval between the adjacent extrema of difference images obeys the geometric distribution and the distribution has periodic peaks for JPEG images. Hence, we adopt the difference image extremum interval histogram and combine the spectral method to obtain the final estimation. The experimental results demonstrate that the proposed detection method outperforms some state-of-the-art methods.

Successful adaptive steganography has mainly focused on embedding the payload while minimizing an appropriately defined distortion function. The application of deep learning to steganalysis has greatly challenged present adaptive steganographic methods, but has also shown the potential for the improvement of steganography. This paper proposes a distortion function generating a framework for steganography. It has three modules: a generator with a U-Net architecture to translate a cover image into an embedding change probability map, a no-pre-training-required double-tanh function to approximate the optimal embedding simulator while preserving gradient norm during backpropagation in the adversarial training, and an enhanced steganalyzer based on a convolution neural network together with multiple high pass filters as the discriminator. Extensive experimental results on different datasets have shown that the proposed framework outperforms the current state-of-the-art steganographic schemes. Moreover, the adversarial training time is reduced dramatically compared with the GAN-based automatic steganographic distortion learning framework (ASDL-GAN).

In this paper, a novel forensics scheme for color image is proposed in color quaternion wavelet transform (CQWT) domain. Compared with discrete wavelet transform (DWT), contourlet wavelet transform, and local binary patterns, CQWT processes a color image as a unit, and so, it can provide more forensics information to identify the photograph (PG) and computer generated (CG) images by considering the quaternion magnitude and phase measures. Meanwhile, two novel quaternion central moments for color images, i.e., quaternion skewness and kurtosis, are proposed to extract forensics features. In the condition of the same statistical model as Farid’s model, the CQWT can boost the performance of the existing identification models. Compared with Farid’s model and Li’s model in 7500 PG and 7500 CG, the quaternion statistical features show a better classification performance. Results in the comparative experiments show that the classification accuracy of the CQWT improves by 19% more than Farid’s model, and the quaternion features approximately improve by 2% more than the traditional.

This paper proposes a novel reversible data hiding (RDH) scheme for medical image based on block classification and code division multiplexing (CDM) in frequency domain. In this scheme, the original medical image is divided into non-overlapping blocks and classified into texture and smooth groups by calculating the mean square error of each block. The texture blocks are then transformed into the frequency domain with integer-to-integer discrete wavelet transform method, and the secret data are embedded into the detail sub-bands of the chosen texture blocks with CDM-based RDH algorithm. Consequently, the detail information of medical image is clarified and the data-embedding capacity is enlarged. Simultaneously, as the spreading sequences employed for data embedding are orthogonal to each other, when the secret message is repeatedly embedded into the cover image, most of the elements of spreading sequences are mutually canceled, which enables the proposed scheme to obtain high visual quality even at large data-embedding capacities. Moreover, on the receiver side, only the receiver with the correct data-embedding and block classification parameters can reconstruct the secret message and the cover image completely; thus, the security of RDH is guaranteed. Experimental results have demonstrated that the proposed scheme can yield better overall performance than other state-of-the-art RDH schemes on medical images.

To improve the prediction accuracy, this paper proposes an adaptive error prediction method based on multiple linear regression (MLR) algorithm. The MLR matrix function that indicates the inner correlations between the pixels and their neighbors is established adaptively according to the consistency of pixels in local area of a natural image, and thus the objected pixel is predicted accurately with the achieved MLR function that denotes the consistency of the neighboring pixels. Compared with the conventional methods that predict the objected pixel with fixed predictors through simple arithmetic combination of its surroundings pixel, the proposed method can provide a comparatively spare prediction-error image for data embedding, and thus can improve the performance of reversible data hiding. Experimental results show that the proposed method outperforms most state-of-the-art error prediction algorithms.

Reversible data hiding algorithm (RDH) has been widely used in multimedia’s copyright protection and content integrity authentication. As a typical RDH scheme, histogram shifting (HS) is extensively investigated due to its high quality of stego-image. Most existing HS based RDH schemes utilize prediction and sorting techniques to build single sharp histogram, which exploit the smooth areas in cover image for data hiding. To take advantages of the correlation among image contents of different texture characteristics, several multiple histograms based RDHs (MH_RDH) are proposed recently, which resort on some rigid rules, e.g. single feature based sorting followed by uniform segmentation of sorted sequence, to construct the multiple histograms. In this paper, the clustering algorithm, i.e. Fuzzy C-means (FCM) clustering, is introduced for the construction of multiple histograms. The FCM equipped with deliberately designed features is employed to classify the cover carriers, e.g. prediction errors, into different clusters with similar traits, which are then used to build the multiple histograms for efficient data embedding. Experimental results demonstrate the superior performance of the proposed scheme over other state-of-the-art ones.

In this paper, based on the block entropy in the spatial domain obtained by summing the squares of all AC coefficients of each 8 × 8 block, a reversible data hiding (RDH) scheme for JPEG images is presented which selects the DCT blocks with minimum entropy for RDH. This scheme applies Parseval theorem and entropy principle to achieve the minimum entropy in the spatial domain. It embeds data into the DCT coefficients based on a histogram–pair RDH scheme using histogram shifting. Moreover, four thresholds including the embedding amplitude threshold, the fluctuation threshold, the lower frequency threshold and the higher frequency threshold, are utilized and adjusted to achieve optimized embedding performance. In particular, the proposed scheme does not embed data into zero-valued DCT coefficients, and can obtain optimal PSNR with relatively small file-size increase. Experimental results have demonstrated that, compared with the prior arts, the quality of proposed marked image is higher, and the file-size increase after RDH is less as well.

Multi-purpose forensics is attracting increasing attention worldwide. In this paper, we propose a multi-purpose method based on densely connected convolutional neural networks (CNNs) for simultaneous detection of 11 different types of image manipulations. An efficient CNN structure has been specifically designed for forensics by considering vital architecture components, including the number of convolutional layers, the size of convolutional kernels, the nonlinear activations, and the type of pooling layer. The dense connectivity pattern, which has better parameter efficiency than the traditional pattern, is explored to strengthen the propagation of features related to image manipulation detection. When compared with four state-of-the-art methods, our experiments demonstrate that the proposed CNN architecture can achieve better performance in multiple operation detections for different image sizes, especially on small image patches. Consequently, the proposed method can accurately detect local image manipulations. The proposed method can achieve better overall performance when tested on different databases as well as better robustness against JPEG compression even under low-quality JPEG compression.

Recently, fully convolutional network (FCN) has been successfully used to locate spliced regions in synthesized images. However, all the existing FCN-based algorithms use real-valued FCN to process each channel separately. As a consequence, they fail to capture the inherent correlation between color channels and the integrity of three channels. So, in this paper, quaternion fully convolutional network (QFCN) is proposed to generalize FCN to quaternion domain by replacing real-valued conventional blocks in FCN with quaternion conventional blocks. In addition, a new color image splicing localization algorithm is proposed by combining QFCNs and superpixel (SP)-enhanced pairwise conditional random field (CRF). QFCNs consider three different versions (QFCN32, QFCN16, and QFCN8) with different up-sampling layers. The SP-enhanced pairwise CRF is used to refine the results of QFCNs. Experimental results on three publicly available datasets demonstrate that the proposed algorithm outperforms the existing algorithms including some conventional algorithms and some deep learning-based algorithms.

Existing video steganography puts much emphasis on the design of algorithms such as mapping rules or distortion functions, thereby ignoring the selection of cover to embed secret information. However, this is just one of the major differences between image steganography and video steganography. In addition, since HEVC is the latest standard in the video codec field, it is of important academic significance and applied value to study HEVC-based steganography. This paper proposes a novel video steganography in HEVC, based on intra-prediction mode (IPM). Firstly, this paper analyzes the probability distribution of 4 × 4 IPMs. Then a cover selection rule combined with the Coding Unit (CU) and Prediction Unit (PU) coding information is proposed, which can improve the security performance of a stego video stream. In addition, matrix coding is used as a coding example to implement the steganography on HEVC video streams. Experimental results show that the proposed algorithm can not only maintain the video quality and the security performance but is also easy to implement. Furthermore, the proposed cover selection rule can also be integrated into other HEVC IPM based steganography.

Nowadays, the most prevailing approach to steganography is the minimal embedding distortion framework, which includes an optimizable distortion function for each cover element and an encoding method to minimize the distortion. With the emergence of Syndrome-Trellis Code, the distortion function plays an increasingly important role in modern adaptive image steganography. In this letter, a new distortion function called generalized uniform embedding distortion (GUED) is proposed for JPEG steganography. The proposed GUED consists of the new distortion measures for both Alternating Current (AC) mode and Discrete Cosine Transform (DCT) block, which are represented in a more general exponential model, aiming to flexibly allocate the embedding data so as to minimize the global changes of the statistics of quantized DCT coefficients after embedding. In addition, an empirical rule is developed to determine the parameters of the exponential function according to the payload and quality factor. By exploring the statistics of both DCT and spatial domains, the proposed GUED is shown to be more consistent with the objective of generalized uniform embedding strategy, i.e., maintaining the relative changes of DCT coefficients to be proportional to their coefficients of variations. Extensive experiments demonstrate that the proposed GUED gains significant performance improvements when compared with its original UERD, and outperforms the state-of-the-art J-UNIWARD with markedly reduced computation time.

Image sharpening is an image enhancement method which has been widely used to improve the quality of images. Therefore, in image forensics, it is required to be identified as all possible manipulations applied in images need to be detected. In recent years, sharpening detection get evolved with new detectors proposed every year to gradually boost the detection performance. This situation continues for several years till the introduction of convolutional neural networks (CNNs). With the assistance of CNNs, the detection of sharpening seems to be completely solved that the detection performance for sharpening achieves perfect, even when the images are weakly sharpened. Is it true that we should no longer pay attention to sharpening forensics any more? To answer this question, in this paper, an anti-forensics method based on generative adversarial network (GAN) is proposed to investigate the philosophy. The images generated via our method possess the feature of sharpening, however, they cannot be simply considered as sharpened images because no traditional sharpening manipulation is applied during the procedure. Observed from the experimental results, even the state-of-art sharpening detector based on CNN can be deceived with the GAN generated images.

With the explosive development in digital techniques, ordinary people without professional training are capable to edit digital images with applications. As a common image processing manipulation, ...

Image sharpening is a basic digital image processing scheme utilized to pursue better image visual quality. From image forensics point of view, revealing the processing history is essential to the content authentication of a given image. Hence, image sharpening detection has attracted increasing attention from researchers. In this paper, a convolutional neural network (CNN) based architecture is reported to detect unsharp masking (USM), the most commonly used sharpening algorithm, applied to digital images. Extensive experiments have been conducted on two benchmark image datasets. The reported results have shown the superiority of the proposed CNN based method over the existed sharpening detection method, i.e., edge perpendicular ternary coding (EPTC).

Dynamic multi-watermarking embeds watermarks into the same region of a cover imperceptibly by simultaneously using different embedding rules. It is regarded as a solution to protect the cover’s copyright or track illegal distributors. Till now, few works have been done on multi-watermarking, especially dynamic multi-watermarking. In this paper, a novel dynamic multi-watermarking rule is proposed, which embeds different watermarks into DWT coefficients controlled by two secret keys, i.e., the sign control key and the factor control key. Two types of the corresponding dynamic multi-watermarking detectors, i.e., optimum and locally optimum, are derived based on the generalized Gaussian distribution and NP criterion. The robustness of the proposed scheme is analyzed using the optimal Lagrange algorithm, which aims to obtain optimal embedding strength factors. Experimental results verify the prominence of the proposed scheme and the validity of theoretical analysis.

A histogram-pair based image reversible data hiding (RDH) scheme is presented in this paper. It embeds data into the prediction-errors generated from a cover image. In doing so, instead of changing a prediction-error x to 2x + b as done in most existing RDH algorithms, where b is a to-be-embedded binary bit, we modify the quantity x to x + b so as to reduce the embedding distortion. Moreover, four thresholds, i.e., the embedding threshold, the fluctuation threshold, the right- and the left-end histogram shrinking thresholds, are employed and adjusted to achieve the highest PSNR at a given embedding rate. The experimental results have demonstrated that, comparing with the state-of-the-arts works, superior performance can be achieved by the proposed method. Particularly, by the proposed histogram shrinking, better embedding result can be achieved for the cover image with high peak bins at both sides of the image spatial histogram (e.g., the JPEG2000 test image Woman).

Recently, several image contrast enhancement methods have been proposed such that the original image can be recovered from its contrast-enhanced version. Hence a flexibility in changing image contrast can be provided when needed. However, the artificial distortions may be introduced into the image content after adopting these methods. Meanwhile, there is lack of using the adequate image quality metrics for performance evaluation. In this paper, a novel reversible data hiding method is proposed for image contrast enhancement. To better preserve image quality, it is restricted that only the adjacent bins in the original image histogram may be merged in the pre-processing. The proposed method has been applied to two image sets and compared with the previous methods. For image quality assessment, the PSNR, SSIM and three no-reference metrics have been adopted in performance evaluation. The experimental results have clearly shown that better visual quality can be achieved with the proposed method. Besides recovering the original images, extra data can be hidden into the contrast-enhanced images and correctly extracted.

In recent years, image sharpening detection has become one of the main topics in the field of image forensics. It is, however, still a challenge to detect the images sharpened with weak sharpening strength. To address this challenge, we propose an efficient method for image sharpening detection. In the proposed method, a ternary coding strategy with adaptive threshold is introduced to reveal the overshoot artifacts caused by weak sharpening. Extensive experiments are conducted to illustrate the superiority of the proposed method. The experimental results show that the proposed method can achieve a considerable improvement in sharpening detection, especially for slightly sharpened images.

The quaternion representation (QR) used in current quaternion-based color image processing creates redundancy when representing a color image of three components by a quaternion matrix having four components. In this paper, both RGB and depth (RGB-D) information are considered to improve QR for efficiently representing RGB-D images. The improved QR fully utilizes the four-dimensional quaternion domain. Using this improved QR, firstly we define the new quaternion pseudo-Zernike moments (NQPZMs) and then propose an efficient computational algorithm for NQPZMs through the conventional pseudo-Zernike moments (PZMs). Finally, we propose an algorithm for color image splicing detection based on the NQPZMs and the quaternion back-propagation neural network (QBPNN). Experimental results on four public datasets (DVMM, CASIA v1.0 and v2.0, Wild Web) demonstrate that the proposed splicing detection algorithm can achieve almost 100% accuracy with the appropriate feature dimensionality and outperforms 14 existing algorithms. Moreover, the comparison of six color spaces (RGB, HSI, HSV, YCbCr, YUV, and YIQ) shows that the proposed algorithm using YCbCr color space has the overall best performance in splicing detection.

Reversible data hiding (RDH) into encrypted images is of increasing attention to researchers as the original content can be perfectly reconstructed after the embedded data are extracted while the content owner’s privacy remains protected. The existing RDH techniques are designed for grayscale images and, therefore, cannot be directly applied to palette images. Since the pixel values in a palette image are not the actual color values, but rather the color indexes, RDH in encrypted palette images is more challenging than that designed for normal image formats. To the best knowledge of the authors, there is no suitable RDH scheme designed for encrypted palette images that has been reported, while palette images have been widely utilized. This has motivated us to design a reliable RDH scheme for encrypted palette images. The proposed method adopts a color partitioning method to use the palette colors to construct a certain number of embeddable color triples, whose indexes are self-embedded into the encrypted image so that a data hider can collect the usable color triples to embed the secret data. For a receiver, the embedded color triples can be determined by verifying a self-embedded check code that enables the receiver to retrieve the embedded data only with the data hiding key. Using the encryption key, the receiver can roughly reconstruct the image content. Experiments have shown that our proposed method has the property that the presented data extraction and image recovery are separable and reversible. Compared with the state-of-the-art works, our proposed method can provide a relatively high data-embedding payload, maintain high peak signal-to-noise ratio values of the decrypted and marked images, and have a low computational complexity.

Image forensics with the presence of an adversary, such as the interplay between the sensor-based camera source identification (CSI) and the fingerprint-copy attack, has attracted increasing attention recently. In this paper, we propose a framework of CSI game with both complete information and incomplete information. A noise level-based counter anti-forensic method is presented to detect the potential fingerprint-copy attack, and unlike the state-of-the-art countermeasure of the triangle test, it does not need to collect the candidate image set. With the existence of countermeasure, a rational forger needs to balance the tradeoff between synthesizing source information and leaving new detectable evidence of raising the noise level of a forged image. The mixed-strategy other than the sequential-move assumption is adopted to solve the games. The Bayesian game is introduced to address the information asymmetry in practice. The Nash equilibrium of both the complete information game and Bayesian game are theoretically analyzed, and the expected Nash equilibrium payoff of a Bayesian game is obtained. Nash equilibrium receiver operating characteristic curves are adopted to evaluate the detection performance. Simulation results show that the information asymmetry can remarkably affect the final detection performance. To our knowledge, this paper is the first attempt in analyzing a Bayesian forensic game with practical information asymmetry.

In this paper, a novel reversible data hiding (RDH) scheme for encrypted digital images using integer wavelet transform, histogram shifting and orthogonal decomposition is presented. This scheme takes advantage of the Laplacian-like distribution of integer wavelet high-frequency coefficients in high frequency sub-bands and the independence of orthogonal coefficients to facilitate data hiding operation in encrypted domain, and to keep the reversibility. Experimental results has demonstrated that this scheme outperforms all of other existing RDH schemes in encrypted domain in terms of higher PSNR at the same amount of payload. Compared with the state-of-the-arts, the proposed scheme can be applied to all natural images with higher embedding rate.

Median filtering, being an order statistic filtering, has been widely used in image denoising and recently also in image anti-forensics and anti-steganalysis. In the past few years, several methods have been developed for median filtering detection. However, it is still a challenging task to detect median filtering in JPEG compressed images. In this paper, we propose a novel method to solve this challenging task. We first generate median filtered residual (MFR), average filtered residual (AFR) and Gaussian filtered residual (GFR) by calculating the differences between an original image and its filtered images. Then, we propose to use two-dimensional autoregressive (2D-AR) model to characterize MFR, AFR and GFR separately, and further combine the 2D-AR coefficients of these three residuals into a set of features. Finally, the extracted feature set is fed into a support vector machine classifier for training and detection. Extensive experiments have demonstrated that compared with existing methods, the proposed one can achieve a considerable improvement in detecting median filtering in heavily compressed images.

Solved the challenge of the privacy-preserving CBIR under cloud environment.Honest-but-curious CSP can retrieve image directly without violating data privacy.The proposed method is based on orthogonal decomposition.Our method can adapt different applications for no restriction of cipher algorithms. In order to protect data privacy, image with sensitive or private information needs to be encrypted before being outsourced to a cloud service provider. However, this causes difficulties in image retrieval and data management. A privacy-preserving content-based image retrieval method based on orthogonal decomposition is proposed in the paper. The image is divided into two different components, for which encryption and feature extraction are executed separately. As a result, cloud server can extract features from an encrypted image directly and compare them with the features of the queried images, so that users can thus obtain the image. Different from other methods, the proposed method has no special requirements to encryption algorithms, which makes it more universal and can be applied in different scenarios. Experimental results prove that the proposed method can achieve better security and better retrieval performance.

In the past more than one decade, hundreds of reversible data hiding (RDH) algorithms have been reported. Via exploring the correlation between the neighboring pixels (or coefficients), extra information can be embedded into the host image reversibly. However, these RDH algorithms cannot be accomplished in encrypted domain directly, since the correlation between the neighboring pixels will disappear after encryption. In order to accomplish RDH in encrypted domain, specific RDH schemes have been designed according to the encryption algorithm utilized. In this paper, we propose a new simple yet effective framework for RDH in encrypted domain. In the proposed framework, the pixels in a plain image are first divided into sub-blocks with the size of $m\times n$ . Then, with an encryption key, a key stream (a stream of random or pseudorandom bits/bytes that are combined with a plaintext message to produce the encrypted message) is generated, and the pixels in the same sub-block are encrypted with the same key stream byte. After the stream encryption, the encrypted $m\times n$ sub-blocks are randomly permutated with a permutation key. Since the correlation between the neighboring pixels in each sub-block can be well preserved in the encrypted domain, most of those previously proposed RDH schemes can be applied to the encrypted image directly. One of the main merits of the proposed framework is that the RDH scheme is independent of the image encryption algorithm. That is, the server manager (or channel administrator) does not need to design a new RDH scheme according to the encryption algorithm that has been conducted by the content owner; instead, he/she can accomplish the data hiding by applying the numerous RDH algorithms previously proposed to the encrypted domain directly.

Current mainstream methods for digital image steganography are content adaptive. That is, the secret messages are embedded in the complicated region in the cover image while minimizing the embedding distortion so as to suppress statistical detectability. Since there is already a practical encoding scheme for data embedding near the payload-distortion bound, the design of the embedding cost function becomes a deterministic part in steganography. Unlike the traditional heuristic hand-crafted method, this paper proposes a novel generative adversarial network based framework to automatically learn the embedding cost function for JPEG steganography. The proposed framework consists of a generator, a gradient-descent friendly inverse discrete cosine transformation module, an embedding simulator and a discriminator for steganalysis. Through training the generator and discriminator in alternation, the embedding cost function can finally be obtained by the trained generator. Experimental results demonstrate that our method can automatically learn a reasonable embedding cost function and achieve a satisfying performance.

In reversible data hiding, the higher embedding capacity and lower distortion are simultaneously expected. Hence, the precise and efficient error-prediction algorithm is essential and crucial. In this paper, a high-performance error-prediction method based on Multiple Linear Regression (MLR) algorithm is proposed to improve the performance of Reversible Data Hiding (RDH). The MLR matrix function that indicates the inner correlations between the pixels and their neighbors is established adaptively according to the consistency of pixels in local area of a natural image, and thus the targeted pixel is predicted accurately with the achieved MLR function that satisfies the consistency of the neighboring pixels. Compared with conventional methods that only predict the targeted pixel with fixed predictors through simple arithmetic combination of its surroundings pixel, the proposed method can provide a sparser prediction-error image for data embedding, and thus improves the performance of RDH. Experimental results have shown that the proposed method outperform the state-of-the-art error prediction algorithms.

Pixel-value-ordering (PVO) is a new technique for the high-fidelity reversible data hiding (RDH). It contains a process of sorting the pixels of a block by their values at first and then embedding into data bits into the maximum or minimum pixels of a block. In this paper, we propose to modify the pixel blocks differently according to how smooth they are, and embed the adequate number of bits into different types of blocks. The pixel blocks are first classified into five types based on the local complexity. The maximum pixels of the most smooth block will be embedded at most four bits, and the less smooth ones are embedded with a lower number of bits. The block classification is dynamically adjusted to achieve the adaptive embedding with the best trade-off between the capacity and the embedding distortion. Experimental results show that the proposed method can give a better performance over the previous PVO-based methods.

An effective feature selection method to capture the weak stego noise is essential to image steganalysis. In the conventional JPEG steganalysis, Gabor filter and DCT filter are both used for residual extraction. However, there are few comparisons in existing convolutional neural networks (CNNs) based JPEG steganalysis using Gabor filter or DCT filter in the pre-processing stage to extract residuals. In this paper, we compare the performance of DCT filter with Gabor filter in the pre-processing phase of the steganalysis CNN. Firstly, we choose the parameters empirically and theoretically for Gabor filters which are used in CNN. Secondly, we improve the performance by removing the ABS layer in the original XuNet. Finally, the experimental results show that using Gabor filters or DCT filter can achieve comparable performance whenever the parameters of pre-processing filters are fixed or learnable. It’s different from the conventional steganalysis method where Gabor filters have advantages over DCT filters. When the parameters of the pre-processing filters are learnable, both Gabor filter and DCT filter can achieve better performance compared with the condition where the parameters are fixed.

Revealing the processing history that a given digital image has gone through is an important topic in digital image forensics. Detection of seam carving, a content-aware image scaling algorithm commonly implemented in commercial image-editing software, has been studied by forensic experts in recent years. In this paper, a convolutional neural network (CNN) architecture is proposed for seam carving detection. Unlike the existing forensic works in detecting seam carving, where the feature selection and the pattern classification are two separated procedures, the proposed CNN-based deep learning architecture learns and then uses more effective features via joint optimization of feature extraction and pattern classification. Experimental results conducted on a large dataset have demonstrated that, compared with the current state-of-the-art, the proposed CNN based deep learning scheme can largely boost the classification rates as the seam carving rate is rather low.

In this paper, a novel set of features based on Quaternion Wavelet Transform (QWT) is proposed for digital image forensics. Compared with Discrete Wavelet Transform (DWT) and Contourlet Wavelet Transform (CWT), QWT produces the parameters, i.e., one magnitude and three angles, which provide more valuable information to distinguish photographic (PG) images and computer generated (CG) images. Some theoretical analysis are done and comparative experiments are made. The corresponding results show that the proposed scheme achieves 18 percents’ improvements on the detection accuracy than Farid’s scheme and 12 percents than Ozparlak’s scheme. It may be the first time to introduce QWT to image forensics, but the improvements are encouraging.

In this paper, a novel code division multiplexing (CDM) algorithm-based reversible data hiding (RDH) scheme is presented. The covert data are denoted by different orthogonal spreading sequences and embedded into the cover image. The original image can be completely recovered after the data have been extracted exactly. The Walsh Hadamard matrix is employed to generate orthogonal spreading sequences, by which the data can be overlappingly embedded without interfering each other, and multilevel data embedding can be utilized to enlarge the embedding capacity. Furthermore, most elements of different spreading sequences are mutually cancelled when they are overlappingly embedded, which maintains the image in good quality even with a high embedding payload. A location-map free method is presented in this paper to save more space for data embedding, and the overflow/underflow problem is solved by shrinking the distribution of the image histogram on both the ends. This would further improve the embedding performance. Experimental results have demonstrated that the CDM-based RDH scheme can achieve the best performance at the moderate-to-high embedding capacity compared with other state-of-the-art schemes.

The local variance of image intensity is a typical measure of image smoothness. It has been extensively used, for example, to measure the visual saliency or to adjust the filtering strength in image processing and analysis. However, to the best of our knowledge, no analytical work has been reported about the effect of JPEG compression on image local variance. In this paper, a theoretical analysis on the variation of local variance caused by JPEG compression is presented. First, the expectation of intensity variance of $8\times 8$ non-overlapping blocks in a JPEG image is derived. The expectation is determined by the Laplacian parameters of the discrete cosine transform coefficient distributions of the original image and the quantization step sizes used in the JPEG compression. Second, some interesting properties that describe the behavior of the local variance under different degrees of JPEG compression are discussed. Finally, both the simulation and the experiments are performed to verify our derivation and discussion. The theoretical analysis presented in this paper provides some new insights into the behavior of local variance under JPEG compression. Moreover, it has the potential to be used in some areas of image processing and analysis, such as image enhancement, image quality assessment, and image filtering.

The embedding is done on encrypted H.264/AVC data without knowing the decryption key.The paired code-word substitution and multiple-base notational system are exploited.Both encryption and data embedding keep the bit-rate intact.Data extraction can be carried out either in encrypted or decrypted domain. Recently, a novel scheme to hide data into encrypted H.264/AVC videos using code-word substitution has been proposed by Xu et al. However, the statistical analysis of CAVLC code-words demonstrate that Xu et al.'s work does not fully exploit redundancy existing in CAVLC code-word for data embedding. In this paper, an improved version of Xu et al.'s data hiding method in encrypted H.264/AVC videos is proposed. Specifically, when suffixLength is equal to 1, data embedding is performed by paired code-word substitution. When suffixLength is greater than 2, not the single code-word substitution but the multiple-based notational system is adopted for data embedding. Experimental results have demonstrated that the improved method is indeed capable of providing a larger embedding capacity in comparison with Xu et al.'s method. Moreover, both encryption and data embedding can be accomplished without affecting the coding efficiency of H.264/AVC by keeping exactly the same bitrate.