Video steganography in the intra prediction mode (IPM) domain embeds secret messages by modifying IPM values. However, such modifications are highly susceptible to detection by video steganalysis techniques, particularly those leveraging recompression-based calibration features. In this paper, the signal restoration phenomenon that occurs during video recompression is first modeled, which reveals the underlying reason for the effectiveness of recompression calibration-based detection features. Based on this insight, a IPM priority-preserving strategy is proposed. This strategy integrates the steganographic modification state with the optimal IPM selection mechanism during recompression, employing dynamic cost revision and joint cost decomposition to guide steganographic modifications toward optimal selection. By aligning modifications with recompression tendency, the proposed method mitigates signal restoration effects, reduces distribution discrepancies in calibration-based detection features, and enhances overall steganographic security. Extensive experimental evaluations demonstrate that the proposed scheme significantly improves resistance against both intra-frame and inter-frame steganalysis features while maintaining superior visual quality and bitrate control.

The rapid advancement of automatic speech recognition (ASR) models has significantly bolstered their multilingual proficiency and robustness, amplifying concerns over user speech privacy. Attackers may use hidden microphones or network attacks to capture and transcribe sensitive user interactions. Whisper, a state-of-the-art (SOTA) multilingual speech recognition model, delivers exceptional transcription accuracy across diverse languages. However, its superior performance also extends privacy leakage risks to multilingual contexts. Previous privacy-preserving methods based on adversarial examples were primarily optimized for monolingual models, limiting their effectiveness in multilingual settings. Moreover, as these perturbation mechanisms were predominantly tailored for English, their transferability to other languages remains constrained. To address this vulnerability, we propose the Disentanglement-based Universal Adversarial Perturbation (DUAP), a privacy-preserving method designed to counteract the Whisper model. Unlike optimization-based approaches, DUAP embeds language-specific features in the latent space to generate robust adversarial perturbations, providing consistent protection across multiple languages and effectively mitigating privacy risks in multilingual contexts. The method employs a two-stage language attack: first, a Language Feature Disentanglement model disentangles and reconstructs language-specific features to produce adversarial examples (AEs); second, gradient-based optimization refines AEs to disrupt Whisper’s language identification module. DUAP’s perturbations, effective in physical and digital settings, achieve SNRs from 40 dB (lightest) to above 17 dB (strongest). Across three Whisper model sizes, DUAP yields WERs over 95% (English), 85% (other languages), and 87% (physical settings), maintaining above 96% under AAC (64, 72 kbps) and MP3 (32, 96 kbps) compressions.

This paper introduces a robust watermarking technique for diffusion model-generated images, which effectively balances watermark robustness, image fidelity, and diversity preservation. Unlike traditional post-hoc approaches, the proposed method embeds watermark information directly into the latent noise of the diffusion model, ensuring seamless integration into the image generation process. This approach minimizes perceptual impact while maintaining high visual quality and diversity of the generated images. Experimental results demonstrate the method’s resilience to various image distortions, including noise, compression, blurring etc., significantly outperforming existing watermarking techniques. The proposed method supports both watermark detection and bit-level extraction, providing a practical solution for secure content protection and traceability in generative models without compromising the integrity of the image generation process.

Combating counterfeit products is crucial for maintaining a healthy market. Recently, Copy Sensitive Graphical Codes (CSGC) have garnered significant attention due to their high sensitivity to illegal physical copying. Copy Detection Patterns (CDP) and Two-Level QR Codes (2LQR code) are two representative methods. CDP offers high efficiency and low cost, enabling use in document authentication and product anti-counterfeiting, and has achieved broad commercial adoption. In contrast, 2LQR code, as a consumer-grade document authentication solution, provides additional private message sharing functionalities. We observe that both the CDP and 2LQR code can be synthesized using textured patterns. To this end, we propose a flexible framework that integrates the stochastic anti-counterfeiting properties of CDP with the private message sharing of 2LQR code. Specifically, we model CDP as a random noise image composed of multiple textured patterns similar to those in 2LQR code, where each pattern represents an informative digit. Thus, both codes can be generated through textured pattern design. We formulate this as a constrained optimization framework called Mixed-Bit Sampling Marking (MSM). The objective incorporates white pixel ratio and spatial randomness, with constraints defined by a flexible modulation function (e.g., DCT or Pearson similarity), customizable to user needs. A two-step sampling algorithm solves the optimization. We demonstrate CDP and 2LQR codes generated via MSM and validate their ability to inherit advantages from both approaches. Experiments show that MSM-generated texture patterns effectively synthesize both CDPs and 2LQR codes, preserving their advantages while offering a novel, flexible solution for document authentication.

With high embedding capacity and security, transform coefficient-based video steganography has become an important branch of video steganography. However, existing steganalysis methods against transform coefficient-based steganography provide insufficient consideration to the prediction process of HEVC compression, which results in steganalysis that is not straightforward and fail to effectively detect adaptive steganography methods in low embedding rate scenarios. In this paper, an HEVC video steganalysis method based on centralized error and attention mechanism against transform coefficient-based steganography is proposed. Firstly, the centralized error phenomenon brought by distortion compensation-based steganography is analyzed, and prediction error maps is constructed for steganalysis to achieve higher SNR(signal-to-noise ratio). Secondly, a video steganalysis network called CESNet (Centralized Error Steganalysis Network) is proposed. The network takes the prediction error maps as input and four types of convolutional modules are designed to adapt to different stages of feature extraction. To address the intra-frame sparsity of adaptive steganography, CEA (Centralized Error Attention) modules based on spatial and channel attention mechanisms are proposed to adaptively enhance the steganographic region. Finally, after extracting the feature vectors of each frame, the detection of steganographic video is completed using the self-attention mechanism. Experimental results show that compared with the existing transform coefficient-based video steganalysis methods, the proposed method can effectively detect multiple transform coefficient-based steganography algorithms and achieve higher detection performance in low payload scenarios.

Copy Detection Pattern (CDP) is a high-density random noise-alike image that exhibits a different noise pattern after physical copying, and is thus treated as a promising anti-counterfeiting solution. However, CDP cannot convey any message, and it is often used in combination with additional carriers, such as QR codes. In this letter, we take the first step towards extending CDP with watermarking functionality. Specifically, we devise a scheme called Mixed-bit Sampling Graphic (MSG), which could realize invisible watermarking and anti-counterfeiting simultaneously. Compared with conventional CDP, the noise pattern generation of MSG is controlled by the portions of sampling over two bit templates. We formulate this mixed-bit sampling process as an optimization problem and solve it using a block coordinate descent sampling algorithm. Experimental results validate that the proposed MSG can effectively communicate watermark bits while retaining the anti-counterfeiting capability of CDP.

HEVC (High Efficiency Video Coding) provides abundant embedding carriers for video steganography, leading to rapid development in the field of video steganography while increasing the urgent demand for video steganalysis. However, existing steganalysis methods against PU (prediction unit) based steganography primarily use the extraction of video statistical features, which ignore the potential information of each frame and fail to effectively detect different PU-based steganography methods. In this paper, a video steganalysis method based on PU maps and multi-scale convolutional residual network is proposed. Firstly, the effects of PU-based steganography on the spatial domain and the compressed domain are analyzed. It is observed that steganography has less impact on the spatial domain, whereas it significantly disrupts the connection between PU blocks in the compressed domain, leaving distinct steganographic traces. Consequently, the PU partition modes containing local connections are introduced to generate PU maps for steganalysis. Secondly, a video steganalysis network called PUSN (Prediction Unit Steganalysis Network) is constructed. The network takes PU maps as input and consists of three parts: feature extraction, feature representation, and binary classification. Additionally, a multi-scale module is proposed to enhance the detection performance. Finally, the detection result of the steganographic video is obtained by the voting mechanism. The experimental results show that compared with the existing steganalysis methods, the proposed method could effectively detect multiple PU-based steganography methods and achieve higher detection accuracy across various embedding rates.

Distortion compensation method is a common way to cope with the distortion drift problem in coefficient domain HEVC steganography. However, it will leave obvious steganographic traces called centralized error (CER). The current coefficient domain HEVC steganography is fragile to CER-based steganalysis. In this article, a novel adaptive HEVC steganography that can resist CER-based steganalysis is proposed. First, the difference of CER between H.264/AVC and HEVC is introduced, and the CER feature in HEVC is re-modeled. Then, from two aspects of overall average distribution and single-frame distribution, we conclude that there is a strong correlation among four components of the CER feature. Last, an adaptive cost function is proposed by maintaining one component distribution to resist steganalysis. Experimental results show that the proposed cost function can effectively improve the security compared with other coefficient-based HEVC steganography. In addition, the proposed steganography outperforms other HEVC steganography in visual quality and bit rate increase.

Modification of the optimal recursive block encoding process is commonly adopted in HEVC steganography based on block partitioning structure to embed secret messages, which inevitably disrupts the optimal rate distortion optimization process, resulting in a degradation of visual quality and an increase in bit rate. In this paper, we analyze the intra frame recursive block encoding process, categorizing modifications based on block partitioning structures into skip-level and non-skip-level modifications. Then, the rate distortion difference between these two types is compared. Additionally, the Maintenance Principle of Quad-tree Structure is introduced, which aims to preserve the stego quad-tree structure as closely as possible to the original one. Furthermore, a new cover mapping method is designed to expand the embedding capacity, and a quad-tree structure-preserving adaptive steganography is proposed. Extensive experimental results demonstrate that the proposed scheme can embed messages with fewer disruptions to the optimal rate distortion optimization process, ultimately improving the visual quality and reducing the bit rate growth.

Deep neural networks have been proven vulnerable to deliberately crafted adversarial example, which cause serious safety and security concerns. Many defense approaches were proposed to resist such threats. However, existing defenses such as pre-compression or adversarial training would degrade the model performance on clean images or incur heavy computational costs. In this work, we propose a plug-and-play defensive module Gradient Sign Inversion (GSI) to defend gradient-based attack. Essentially, GSI attempts to inverse the direction of the backpropagated gradient for the victim model, disturbing the adversarial example generation of the attacking while retaining the performance of the vanilla network on genuine inputs. Specifically, an additive model based on periodic trigonometric function is established by investigating the necessary conditions that a suitable defensive module should have. By enforcing constraints on the defensive module, the parameters of GSI are determined, accompanied by a theoretical justification. Interestingly, we observe that the proposed GSI not only prevents the gradient-based adversarial attack, but can even improve the confidence of the ground-truth label when initiating an attack, making the attack betray as a defense. Source code is publicly available at https://github.com/JidaDiao/GSI.

Optical Character Recognition (OCR) enables automatic text extraction from scanned or digitized text images, but it also makes it easy to pirate valuable or sensitive text from these images. Previous methods to prevent OCR piracy by distorting characters in text images are impractical in real-world scenarios, as pirates can capture arbitrary portions of the text images, rendering the defenses ineffective. In this work, we propose a novel and effective defense mechanism termed the Universal Defensive Underpainting Patch (UDUP) that modifies the underpainting of text images instead of the characters. UDUP is created through an iterative optimization process to craft a small, fixed-size defensive patch that can generate non-overlapping underpainting for text images of any size. Experimental results show that UDUP effectively defends against unauthorized OCR under the setting of any screenshot range or complex image background. It is agnostic to the content, size, colors, and languages of characters, and is robust to typical image operations such as scaling and compressing. In addition, the transferability of UDUP is demonstrated by evading several off-the-shelf OCRs. The code is available at https://github.com/QRICKDD/UDUP.

Digital documents, as a twin of hard copy, are increasingly being used as credible evidence. Unfortunately, digital document images easily suffer forgery or malicious manipulation, with the availability of sophisticated image editing tools. To verify and detect the possible forgeries for a given document, a number of forensic schemes have been developed. However, in the real-world scenario, the doctored image could be further processed or transmitted over a channel with unknown distortion, which dramatically degrade the forgery detection performance. In this work, we make the first step towards designing a robust document image forgery localization against image blending. Specifically, we propose an encoder-decoder neural network architecture consisting of three modules. The first module is responsible for capturing the multi-scale features from the high-level feature maps, and the remaining two attention-based modules aim to extract low-level local features and high-level global features. For training the model, we construct a dedicated forgery document database processed by several recent image blending procedures. Extensive experiments demonstrate the effectiveness and superiority of the proposed method in detecting the forgery that undergoes image blending. The source code, models and the constructed image dataset are publicly available at https://github.com/lwp0201/Image-Forgery-Localization-Against-Image-Blending.

Despite neural network-based speaker recognition systems (SRS) have enjoyed significant success, they are proved to be quite vulnerable to adversarial examples. In practice, the SRS model parameters are not always available. Attackers have to probe the model only via querying, and such decision-based attacking merely relies on the output label is quite challenging. This letter proposes a two-step query-efficient decision-based attack based on local low-frequency perturbation. Specifically, instead of imposing perturbation on the entire audio sample, a local attacking region is firstly sought, confining the perturbed distortion to a local region. Second, considering that the majority of energy concentrates on the low-frequency bands, the proposed method suggests performing perturbation generation in the low-frequency domain. Experimental results demonstrate that, compared with the recent methods, our method could implement target attacking to SRS with a higher attacking success rate, at the cost of much lower queries and adversarial perturbation.

Nowadays, the convolutional neural network (CNN) based steganalysis has achieved remarkable performance in the well-controlled lab environment. However, the cover source mismatch (CSM) problem, which can be attributed to the discrepancy between the training, and evaluation datasets, is still one of the pivotal obstacles for adapting the steganalysis into real-world applications. In this letter, we propose to merge the domain adaptation strategy into CNN-based audio steganalysis for handling the CSM problem. Specifically, the proposed framework contains three components: feature extractor, steganalytic classifier, and domain discriminator. The cascade of feature extractor, and steganalytic classifier compose the typical supervised steganalysis model. The unsupervised domain adaptation is implemented by the domain adversarial training between the feature extractor, and domain discriminator. Ultimately, the feature extractor is trained to extract the steganalytic, and domain-invariant features. It aims to reduce the domain gap between the training data, and testing data. The experimental results show that our approach could effectively mitigate the CSM impact caused by the diversity of audio recording devices.

In the field of steganography, to effectively hide the secret message, it is of great importance to determine which part of the steganographic cover is suitable for embedding. Currently, most of the existing works focus on the image cover, while few works touch the audio cover case. In this work, we attempt to characterize the complexity of audio for selecting the steganographic cover. Specifically, the original cover is first convoluted with a specially designed adaptive convolution kernel. Based on the residual between the original and the convoluted audio, we derive a quantity for measuring the complexity of each frame for a given audio clip. Experimental results verify the usability of the proposed complexity indicator, suggesting high-complexity audio cover is favorable for data embedding. It is also found that the proposed complexity indicator could further boost the steganographic performance of the state-of the-art audio steganography methods. The source code is publicly available at https://github.com/capzxy/audio-complexity.

Adaptive multirate (AMR) compression audio has been exploited as an effective forensic evidence to justify audio authenticity. Little consideration has been given, however, to antiforensic techniques capable of fooling AMR compression forensic algorithms. In this paper, we present an antiforensic method based on generative adversarial network (GAN) to attack AMR compression detectors. The GAN framework is utilized to modify double AMR compressed audio to have the underlying statistics of single compressed one. Three state-of-the-art detectors of AMR compression are selected as the targets to be attacked. The experimental results demonstrate that the proposed method is capable of removing the forensically detectable artifacts of AMR compression under various ratios with an average successful attack rate about 94.75%, which means the modified audios generated by our well-trained generator can treat the forensic detector effectively. Moreover, we show that the perceptual quality of the generated AMR audio is well preserved.

Deep learning, especially the convolutional neural network (CNN), has enjoyed significant success in many fields, e.g., image recognition. Recently, CNN has successfully applied to multimedia steganalysis. However, the detection performance is still unsatisfactory. In this work, we propose an improved CNN-based method for audio steganalysis. Specifically, a special convolutional layer is first carefully designed, which could capture the minor steganographic noise. Then, a truncated linear unit is adapted to activate the output of shallow convolutional layer. In addition, we employ the average pooling to minimize the over-fitting risk. Finally, a parameter transfer strategy is adopted, aiming to boost the detection performance for the low embedding-rate cases. The experimental results evaluated on 30,000 audio clips verify the effectiveness of our method for a variety of embedding rates. Compared with the existing CNN-based steganalysis methods, our proposed method could achieve superior performance. To facilitate the reproducible research, the source code will be released at GitHub.

With the advent of sophisticated and low‐cost video editing software, digital videos are highly vulnerable to be tampered. The authenticity and integrity identification of digital videos is an urgent issue. In this study, an effective method to detect double High Efficiency Video Coding (HEVC) video compression with different quantisation parameter (QP) is proposed, which often occurs in the video tampering process. First, the effects of QP on the distributions of Discrete Cosine Transform (DCT) coefficients and Transform Unit (TU) size are analysed. Then a feature set including 17 features is derived from quantised DCT coefficients and TU size. It can characterize the statistical differences between single and double compressed videos. Finally, the Library for Support Vector Machine classifier is exploited to identify whether a given HEVC video has been double compressed or not. Experimental results demonstrate that the authors’ detection method has a good comprehensive performance.

An efficient method of completely separable reversible data hiding in encrypted images is proposed. The cover image is first partitioned into nonoverlapping blocks and specific encryption is applied to obtain the encrypted image. Then, image difference in the encrypted domain can be calculated based on the homomorphic property of the cryptosystem. The data hider, who does not know the original image content, may reversibly embed secret data into image difference based on two-dimensional difference histogram modification. Data extraction is completely separable from image decryption; that is, data extraction can be done either in the encrypted domain or in the decrypted domain, so that it can be applied to different application scenarios. In addition, data extraction and image recovery are free of any error. Experimental results demonstrate the feasibility and efficiency of the proposed scheme.

Assessing the visual realism of images is increasingly becoming an essential aspect of fields ranging from computer graphics (CG) rendering to photo manipulation. In this paper we systematically evaluate factors underlying human perception of visual realism and use that information to create an automated assessment of visual realism. We make the following unique contributions. First, we established a benchmark dataset of images with empirically determined visual realism scores. Second, we identified attributes potentially related to image realism, and used correlational techniques to determine that realism was most related to image naturalness, familiarity, aesthetics, and semantics. Third, we created an attributes-motivated, automated computational model that estimated image visual realism quantitatively. Using human assessment as a benchmark, the model was below human performance, but outperformed other state-of-the-art algorithms.

Computer-generated (CG) face images are common in video games, advertisements, and other media. CG faces vary in their degree of realism, a factor that impacts viewer reactions. Therefore, efficient control of visual realism of face images is important. Efficient control is enabled by a deep understanding of visual realism perception: the extent to which viewers judge an image as a real photograph rather than a CG image. Across two experiments, we explored the processes involved in visual realism perception of face images. In Experiment 1, participants made visual realism judgments on original face images, inverted face images, and images of faces that had the top and bottom halves misaligned. In Experiment 2, participants made visual realism judgments on original face images, scrambled faces, and images that showed different parts of faces. Our findings indicate that both holistic and piecemeal processing are involved in visual realism perception of faces, with holistic processing becoming more dominant when resolution is lower. Our results also suggest that shading information is more important than color for holistic processing, and that inversion makes visual realism judgments harder for realistic images but not for unrealistic images. Furthermore, we found that eyes are the most influential face part for visual realism, and face context is critical for evaluating realism of face parts. To the best of our knowledge, this work is a first realism-centric study attempting to bridge the human perception of visual realism on face images with general face perception tasks.

Digital video sometimes needs to be stored and processed in an encrypted format to maintain security and privacy. For the purpose of content notation and/or tampering detection, it is necessary to perform data hiding in these encrypted videos. In this way, data hiding in encrypted domain without decryption preserves the confidentiality of the content. In addition, it is more efficient without decryption followed by data hiding and re-encryption. In this paper, a novel scheme of data hiding directly in the encrypted version of H.264/AVC video stream is proposed, which includes the following three parts, i.e., H.264/AVC video encryption, data embedding, and data extraction. By analyzing the property of H.264/AVC codec, the codewords of intraprediction modes, the codewords of motion vector differences, and the codewords of residual coefficients are encrypted with stream ciphers. Then, a data hider may embed additional data in the encrypted domain by using codeword substitution technique, without knowing the original video content. In order to adapt to different application scenarios, data extraction can be done either in the encrypted domain or in the decrypted domain. Furthermore, video file size is strictly preserved even after encryption and data embedding. Experimental results have demonstrated the feasibility and efficiency of the proposed scheme.

Petitcolas has proposed a steganographic technique called MP3Stego which can hide secret messages in a MP3 audio. This technique is well-known because of its high capacity. However, in rare cases, the normal audio encoding process will be terminated due to the endless loop problem caused by embedding operation. In addition, the statistical undetectability of MP3Stego can be further improved. Inspired by MP3Stego, a new steganographic method for MP3 audio is proposed in this paper. The parity bit of quantization step rather than the parity bit of block size in MP3Stego is employed to embed secret messages. Compared with MP3Stego, the proposed method can avoid the endless loop problem and achieve better imperceptibility and higher security.

Moving shadows cause serious problem while segmenting and extracting foreground from video sequences, due to the misclassification of moving shadow as foreground. In order to detect the moving objects accurately, this paper proposes to remove the moving shadows based on the similarity between little textured patches. The potential shadow points are detected by analyzing the intensity and color properties firstly, and then, the shadow detection approach is improved by evaluating the textural similarity between the current frame and the corresponding background model. Experiment results on both indoor and outdoor scenes exhibit that the proposed method succeeds in removing shadows robustly and achieves the real-time performance.

The multimedia bitstream is comprised of variable length codes (VLC) to a large extent. If data hiding is carried out in multimedia compressed domain, VLC are usually chosen as the hidden object. The VLC-coded efficiency is a key factor which can directly affect the method of data hiding. MPEG-I layer III (MP3) audio is chosen as analytical object in this paper. The VLC-coded rule is obtained by analyzing the VLC-coded efficiency of the MP3 audios. The process of variable length coding in MP3 audio can be understood further passing through the VLC-coded rule. At the same time, it also provides a favorable theoretical base for seeking for effective data hiding methods in MP3 audio compressed domain.

This work proposes an audio zero-watermark algorithm which combined DCT and Zernike Moments. The proposed algorithm combines a part of samples and low-frequency DCT coefficients of the same volume to perform low-order Zernike transform, and then construct watermark based on Zernike moments. Experimental results show that the proposed algorithm can resist both common signal processing and geometrical attacks effectively.

Ray-space representation is the main technology to realize Free Viewpoint Video (FVV) system with complicated scene. Ray-space data consists of various lines with different direction. Ray-space interpolation and compression are two key techniques to be solved. In this paper, correlations between multiple epipolar lines in the Ray-space data is analyzed, and a new algorithm of Ray- Space interpolation with multiepipolar lines matching is proposed. Experimental results show that the proposed scheme achieves higher PSNR than the pixel-based matching interpolation method and the block-based matching interpolation method in interpolating the ray-space data and rendering arbitrary viewpoint image.

An audio watermarking scheme based on wavelet packet and psychoacoustic model is presented. Wavelet packet is a very good tool to analysis the audio signal which is non-stationary. The algorithm has better imperceptibility by using masking effect in human auditory system. The masking threshold can be computed in wavelet domain. Thus the computational complexity is reduced greatly, because it doesn’t like MPEG algorithm which should perform FFT. The watermark can be blind extracted by using linear predictive coding. Experimental results show that the watermark is imperceptible and the algorithm is robust to many attacks, such as mp3 compression, noise addition, requantization, low pass filtering, D/A -A/D and so on.

There are two problems in video communication, one is related to heterogeneity of networks, and the other involves reliability of transmission. Layered coding is designed to solve client heterogeneity problems, and multiple description coding is an effective method for robust transmission. Multiple description layered coding (MDLC) of video combines advantages of layered coding and multiple description coding. In this paper, a new MDLC scheme of video sequence is proposed based on macroblock splitting technique. In addition, other three schemes of MDLC are also given based on row-, column-, and framedecomposition. Experimental results show that the proposed MDLC scheme with macroblock splitting (MDLC-MS) has advantages in adaptability of network heterogeneity and transmission reliability.