Zhenxing Qian

Deep learning-based video steganography has made significant strides, yet conventional explicit methods often suffer from cover distortion and reduced extraction accuracy at high capacities. In this paper, we propose an implicit video steganography framework that treats video hiding and recovery as a dual-stream generation process leveraging implicit neural representations. Instead of altering existing carriers, secret information is encoded within the neural network’s weights, making it an inherent part of the generation process. We introduce a dual-stream input encoding mechanism that decouples the input space into temporal and cryptographic encodings to ensure covert transmission, allowing only authorized receivers to recover hidden content. Furthermore, a multi-scale generation network, incorporating frequency-aware upscaling and statistical distribution loss, is presented to achieve high-quality reconstruction. Extensive experiments demonstrate that our approach achieves state-of-the-art results, minimizing detectable discrepancies while concealing up to seven secret videos within a single carrier. Our method significantly outperforms existing benchmarks by a margin of over 10 dB in peak signal-to-noise ratio, highlighting its superior imperceptibility, accuracy, and security.

We propose a sorting-and-prediction integrated framework for high-capacity reversible data hiding (RDH). The framework constructs a two-dimensional sorting criterion that jointly leverages predicted pixel values and local complexity. A snake-like scanning strategy is then applied to reorganize pixels into a globally smooth embedding domain while maintaining full reversibility. This design suppresses modification propagation in textured regions and stabilizes embedding performance under large payloads. On this basis, a four-layer LPVO-based embedding scheme is developed to further enhance capacity while preserving image fidelity. Experiments are conducted on six standard images and the Kodak dataset. Results show that the proposed method consistently outperforms representative RDH approaches across varying embedding capacities. On the Kodak dataset, the proposed method achieves an average PSNR of 60.43 dB at 20,000 bits, outperforming a recent RL-based RDH method by 0.59 dB. These results validate that constructing a globally smooth embedding domain is an effective strategy for high-fidelity, high-capacity RDH.

Steganography can be mainly classified into two categories: modification-based and generation-based approaches. Modification-based steganography embeds secret messages by altering specific statistical properties of the cover image, offering relatively high capacity and robustness. However, such modifications also make the stego images more susceptible to be detected by steganalysis tools. In contrast, generation-based steganography, which directly generates them via generative models, provides higher security but suffers from limited hiding capacity and poor robustness. To address these limitations, we propose a unified deep network for image steganography (U-StegNet) that integrates the high capacity and robustness of modification-based steganography with the high security of generation-based steganography. Specifically, we design a dual stage hiding strategy. In the first stage, we utilize a generative model to produce an intermediate image that hides part of the secret message while preserving statistical properties consistent with natural images, thereby enhancing resistance to steganalysis. In the second stage, we design a U-Net-based encoder to hide the remaining secret message into the intermediate image, while simultaneously enhancing fine-grained visual details. Moreover, we introduce a message division module that adaptively allocates the message length for each stage based on the image content and capacity constraints, thereby improving both imperceptibility and robustness. Extensive experimental results demonstrate the effectiveness of our U-StegNet. It achieves a PSNR of 44.41 dB and an SSIM of 0.9968 on stego images, which demonstrates the high imperceptibility of our method. Furthermore, it maintains strong robustness across nine types of common attacks, with message extraction accuracy nearly above 98%.

Diffusion models have made tremendous progress in generating visually realistic images. However, these images are statistically different from the real images, which could be accurately classified by carefully designed detectors. To evade detection, researchers have proposed various adversarial example generation schemes for AI-generated images. Despite the progress, most of these schemes have the tendency to post-process the images and the distortion is inevitable. In this paper, we propose an Adversarial Diffusion Model (ADM), which is able to directly generate high quality and undetectable images from scratch on top of a pre-trained stable diffusion model. In the ADM, an adversarial denoising U-Net is proposed for searching an adversarial latent. This latent is helpful for generating a prompt consistent adversarial example which is able to deceive the detector. Then, we propose a latent compensation module to make the adversarial examples have a similar reconstruction error to that of the real images. We further propose an adversarial decoder to minimize the difference between the high-frequency components of the real and adversarial examples. Comprehensive experiments are carried out to demonstrate the advantages of our ADM in generating adversarial AI-generated images.

The proliferation of AI-generated imagery poses escalating challenges for multimedia forensics, yet many existing detectors depend on assumptions about the internals of specific generative models, limiting their cross-model applicability. We introduce a self-supervised approach for detecting AI-generated images that leverages camera metadata—specifically exchangeable image file format (EXIF) tags—to learn features intrinsic to digital photography. Our pretext task trains a feature extractor solely on camera-captured photographs by classifying categorical EXIF tags (e.g., camera model and scene type) and pairwise-ranking ordinal and continuous EXIF tags (e.g., focal length and aperture value). Using these EXIF-induced features, we first perform one-class detection by modeling the distribution of photographic images with a Gaussian mixture model and flagging low-likelihood samples as AI-generated. We then extend to binary detection that treats the learned extractor as a strong regularizer for a classifier of the same architecture, operating on high-frequency residuals from spatially scrambled patches. Extensive experiments across various generative models demonstrate that our EXIF-induced detectors substantially advance the state of the art, delivering strong generalization to in-the-wild samples and robustness to common benign image perturbations.