神经网络模型隐写可用于版权保护与隐蔽通信，但现有方法大多依赖模型中间层参数，导致嵌入容量有限、抗噪声能力不足。为此，本文提出一种多层级联动态嵌入的隐写方案。首先设计了多层权重分配函数，利用动态权重将秘密数据自适应地分配并嵌入到多层神经网络中，在扩展嵌入容量的同时提升了鲁棒性；然后，在模型中引入通道注意力模块，通过注意力特征强化抵消秘密数据嵌入对模型整体性能的影响。大量仿真实验表明，所提方案可以高效地解决数据隐藏的场景问题，在嵌入容量、鲁棒性、安全性方面均优于现有方法。
Neural network steganography holds potential for copyright protection and covert communication. However, most existing methods rely on intermediate layer parameters of the model, which often leads to limited embedding capacity and insufficient robustness against noise. To address these issues, this paper proposes a steganographic scheme based on multi-level cascaded dynamic embedding. Firstly, a multi-layer weight distribution function is designed to adaptively allocate and embed secret data across multiple layers of the neural network using dynamic weights, thereby expanding the embedding capacity and enhancing robustness. Furthermore, a channel attention module is introduced into the model to counteract the performance degradation caused by secret data embedding, leveraging enhanced attention features to balance information embedding with model functionality preservation. Extensive simulation experiments demonstrate that the proposed scheme effectively addresses data hiding challenges and outperforms existing methods in terms of embedding capacity, robustness, and security.
[1] 秦川, 董腾林, 姚恒. 基于风格迁移纹理合成与识别的构造式信息隐藏[J]. 软件学报, 2023, 34(12): 5773-5786. Qin C, Dong T L, Yao H. Constructive data hiding based on texture synthesis and recognition with image style transfer [J]. Journal of Software, 2023, 34(12): 5773-5786. (in Chinese)
[2] 袁超, 王宏霞, 何沛松. 基于注意力机制的高容量通用图像隐写模型[J]. 软件学报, 2024, 35(3): 1502-1514. Yuan C, Wang H X, He P S. High-capacity universal image steganographic model based on attention mechanism [J]. Journal of Software, 2024, 35(3): 1502-1514. (in Chinese)
[3] 王子驰, 李斌, 冯国瑞, 等. 数字图像隐写分析综述[J]. 应用科学学报, 2024, 42(5): 723-732. Wang Z C, Li B, Feng G R, et al. Review of steganalysis for digital images [J]. Journal of Applied Sciences, 2024, 42(5): 723-732. (in Chinese)
[4] 段新涛, 白鹿伟, 徐凯欧, 等. 基于卷积神经网络的轻量高效图像隐写[J]. 应用科学学报, 2025, 43(1): 80-93. Duan X T, Bai L W, Xu K O, et al. Lightweight and efficient image steganography based on convolutional neural network [J]. Journal of Applied Sciences, 2025, 43(1): 80-93. (in Chinese)
[5] Uchida Y, Nagai Y, Sakazawa S, et al. Embedding watermarks into deep neural networks [C]//2017 ACM International Conference on Multimedia Retrieval. ACM, 2017: 269-277.
[6] Adi Y, Baum C, Cisse M, et al. Turning your weakness into a strength: watermarking deep neural networks by backdooring [C]//27th USENIX Security Symposium (USENIX Security 18), 2018: 1615-1631.
[7] Salem A, Backes M, Zhang Y. Get a model! Model hijacking attack against machine learning models [C]//29th ACM Conference on Computer and Communications Security (CCS’22). ACM, 2022: 3089-3104.
[8] Wu H Z, Gen L, Yao Y W, et al. Watermarking neural networks with watermarked images [J]. IEEE Transactions on Circuits and Systems for Video Technology, 2021, 31(7): 2591-2601.
[9] Yang Z Y, Wang Z C, Zhang X P. A general steganographic framework for neural network models [J]. Information Sciences, 2023, 643: 119250.
[10] Wang J F, Wu H Z, Zhang X P, et al. Watermarking in deep neural networks via error back-propagation [C]//Media Watermarking, Security, and Forensics Conference, 2020.
[11] Wang Z C, Feng G R, Wu H Z, et al. Data hiding in neural networks for multiple receivers [J]. IEEE Computational Intelligence Magazine, 2021, 16: 70-84.
[12] Wang Z Y. Multi-source data hiding in digital images [J]. Symmetry, 2022, 14(5): 890.
[13] Yang Z Y, Wang Z C, Zhang X P, et al. Multi-source data hiding in neural networks [C]//24th International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2022: 1-6.
