Jianhua Yang

Image hiding fully explores the hidden potential of deep learning-based models, aiming to conceal image-level messages within cover images and reveal them from stego images to achieve covert communication. Although image hiding has made significant progress, achieving imperceptible color JPEG image hiding remains challenging. Existing hiding schemes are easily detected by the naked eyes or steganalyzers due to the cover type confined to the spatial domain, single-range feature extraction and attacks, and insufficient loss constraints. In this paper, we propose a multi-range representations-driven adversarial stego generation framework called MRAG for color JPEG image hiding. The proposed MRAG can address the aforementioned shortcomings within a unified framework, enabling steganalysis imperceptibility in color JPEG image hiding while ensuring secret restorability and visual imperceptibility. Specifically, MRAG integrates the local-range characteristic of the convolution and the global-range modeling of the transformer. This design stems from the fact that steganalyzers typically combine local-range and global-range information to better capture hidden traces. Meanwhile, we introduce a loss function, termed the features angle-norm disentanglement loss, to launch multi-range representations-driven feature-level adversarial attacks. It computes the adversarial loss between covers and stegos based on the surrogate steganalyzer’s classified features, i.e., the features before the last fully connected layer. Under the dual constraints of features angle and norm, MRAG can delicately encode the concatenation of cover and secret into subtle adversarial perturbations from local and global ranges relevant to steganalysis. Therefore, the resulting stego can achieve visual and steganalysis imperceptibility. Moreover, coarse-grained and fine-grained frequency decomposition operations are devised to transform the input, introducing multi-grained frequency information. Extensive experiments demonstrate that MRAG can achieve state-of-the-art performance.

Image forgery localization is critical in defending against the malicious manipulation of image content, and is attracting increasing attention worldwide. In this paper, we propose a Dual-domain Fusion Swin Transformer U-Net (DFST-UNet) for image forgery localization. DFST-UNet is built on a U-shaped encoder–decoder architecture. Swin Transformer blocks are integrated into the U-Net architecture to capture long-range context information and perceive forged regions at different scales. Considering the fact that high-frequency forgery information is an essential clue for forgery localization, a dual-stream encoder is proposed to comprehensively expose forgery clues in both the RGB domain and the frequency domain. A novel high-frequency feature extractor module (HFEM) is designed to extract robust high-frequency features. A hierarchical attention fusion module (HAFM) is designed to effectively fuse the dual-domain features. Extensive experimental results demonstrate the superiority of DFST-UNet over the state-of-the-art methods in the task of image forgery localization.

With the rapid development of image generation techniques, it becomes much more difficult to distinguish high‐quality computer‐generated (CG) images from photographic (PG) images, challenging the authenticity and credibility of digital images. Therefore, distinguishing CG images from PG images has become an important research problem in image forensics, and it is crucial to develop reliable methods to detect CG images in practical scenarios. In this paper, we proposed a forensics contrastive learning (FCL) framework to adaptively learn intrinsic forensics features for the general and robust detection of CG images. The data augmentation module is specially designed for CG image forensics, which reduces the interference of forensic‐irrelevant information and enhances discrimination features between CG and PG images in both the spatial and frequency domains. Instance‐wise contrastive loss and patch‐wise contrastive loss are simultaneously applied to capture critical discrepancies between CG and PG images from global and local views. Extensive experiments on different public datasets and common postprocessing operations demonstrate that our approach can achieve significantly better generalization and robustness than the state‐of‐the‐art approaches. This manuscript was submitted as a pre‐print in the following link https://papers.ssrn.com/-sol3/papers.cfm?abstract_id=4778441 .

A great challenge to steganography has arisen with the wide application of steganalysis methods based on convolutional neural networks (CNNs). To this end, embedding cost learning frameworks based on generative adversarial networks (GANs) has been proposed and achieved success for spatial image steganography. However, the application of GAN to JPEG steganography is still in the prototype stage; its antidetectability and training efficiency should be improved. In conventional steganography, research has shown that the side information calculated from the precover can be used to enhance security. However, it is hard to calculate the side information without the spatial domain image. In this work, an embedding cost learning framework for JPEG image steganography via a GAN (JS–GAN) has been proposed, the learned embedding cost can be further adjusted asymmetrically according to the estimated side information (ESI). Experimental results have demonstrated that the proposed method can automatically learn a content‐adaptive embedding cost function, and using the ESI properly can effectively improve the security performance. For example, under the attack of a classic steganalyzer GFR with a quality factor of 75 and 0.4 bpnzAC, the proposed JS–GAN can increase the detection error by 2.58% over J‐UNIWARD, and the ESI–aided version JS–GAN (ESI) can further increase the security performance by 11.25% over JS–GAN.

Steganography is a technique that hides secret messages into a public multimedia object without raising suspicion from third parties. However, most existing works cannot provide good robustness against lossy JPEG compression while maintaining a relatively large embedding capacity. This paper presents an end-to-end robust steganography system based on the invertible neural network (INN). Instead of hiding in the spatial domain, our method directly hides secret messages into the discrete cosine transform (DCT) coefficients of the cover image, which significantly improves the robustness and anti-steganalysis security. A mutual information loss is first proposed to constrain the flow of information in INN. Besides, a two-way fusion module (TWFM) is implemented, utilizing spatial and DCT domain features as auxiliary information to facilitate message extraction. These two designs aid in recovering secret messages from the DCT coefficients losslessly. Experimental results demonstrate that our method yields significantly lower error rates than other existing hiding methods. For example, our method achieves reliable extraction with 0 error rate for 1 bit per pixel (bpp) embedding payload; and under the JPEG compression with quality factor QF=10, the error rate of our method is about 22% lower than the state-of-the-art robust image hiding methods, which demonstrates remarkable robustness against JPEG compression.

JPEG steganography has become a research hotspot in the field of information hiding. However, the capacity of conventional JPEG steganography methods is hard to meet the requirements in high-capacity application scenarios and also can not extract secret messages accurately after JPEG compression. To mitigate these problems, we propose a high-capacity and robust JPEG steganography based on adversarial training called HRJS, which implements an end-to-end framework in the JPEG domain for the first time. The encoder is responsible for embedding the secret message while the decoder can reconstruct the original secret message. To enhance robustness, an attack module forces the neural network to automatically learn how to correctly recover the secret message after an attack. Experimental results show that our method achieves near 100 decoding accuracy against JPEG_50 compression at 1/3 bits per channel (bpc) payload while preserving the imperceptibility of the stego image. Compared with conventional JPEG steganography methods, the proposed method is feasible with high capacity (e.g., 1 bpc) and has an obvious advantage in terms of robustness against JPEG compression at the same time.