基于非局部机制的可逆神经网络视频隐写研究

doi:10.3969/j.issn.1671-1122.2026.01.012

摘要/Abstract

摘要：

文章设计了一种可逆神经网络视频隐写模型。该模型通过优化设计的非局部机制函数，替代当前可逆神经网络中普遍应用的稠密型连接函数，有效解决了视频载体中提取出的秘密信息质量不高的问题；通过设计叠加态的加密和解密结构，提升了模型的安全性；通过优化可逆神经网络块的数量，提高了网络效率。实验结果表明，在利用可逆神经网络进行视频隐写时，与稠密型网络结构相比，采用非局部机制的可逆神经网络能够恢复更高质量的秘密信息，且嵌入秘密后视频的失真更少。同时，使用叠加态的加解密结构，在应用层面显著提升了可逆神经网络视频隐写的安全性。

关键词: 视频隐写, 可逆神经网络, 非局部机制

Abstract:

The article proposed a novel reversible neural network-based video steganography model. By replacing the widely used dense connection functions in current reversible neural networks with an optimized non-local mechanism function, the model effectively addressed the issue of low-quality secret information extraction from video carriers. Through the design of a superposition-based encryption and decryption structure, the model’s security was enhanced. Additionally, by optimizing the number of reversible neural network blocks, the network efficiency was improved. Experimental results demonstrate that, when utilizing reversible neural networks for video steganography, compared to dense network structures, the reversible neural network employing the nonlocal mechanism recovers secret information of higher quality, and the video distortion after secret embedding is reduced. Furthermore, the use of the superposition-based encryption and decryption structure significantly enhances the security of reversible neural network-based video steganography at the application level.

Key words: video steganography, reversible neural network, nonlocal mechanism

中图分类号:

TP309

钮可, 胡方锰, 李军. 基于非局部机制的可逆神经网络视频隐写研究[J]. 信息网络安全, 2026, 26(1): 139-149.

NIU Ke, HU Fangmeng, LI Jun. Research on Reversible Neural Network Video Steganography Based on Nonlocal Mechanism[J]. Netinfo Security, 2026, 26(1): 139-149.

图/表 10

图1

图2

图3

图4

图5

表1

表2

表3

表4

表5

参考文献 32

[1]	BALUJA S. Hiding Images in Plain Sight: Deep Steganography[EB/OL]. (2017-12-04)[2025-07-06]. https://www.semanticscholar.org/paper/Hiding-Images-in-Plain-Sight%3A-Deep-Steganography-Baluja/d511567a87218a555a4cc662c42ee89f044a5852.
[2]	HAYES J, DANEZIS G. Generating Steganographic Images via Adversarial Training[EB/OL]. (2017-03-01)[2025-07-06]. https://www.semanticscholar.org/paper/Generating-steganographic-images-via-adversarial-Hayes-Danezis/2da1a80955df1612766ffdf63916a6a374780161.
[3]	KUNHOTH J, SUBRAMANIAN N, AL-MAADEED S, et al. Video Steganography: Recent Advances and Challenges[J]. Multimedia Tools and Applications, 2023, 82(27): 41943-41985. doi: 10.1007/s11042-023-14844-w
[4]	CHAN C K, CHENG L M. Hiding Data in Images by Simple LSB Substitution[J]. Pattern Recognition, 2004, 37(3): 469-474. doi: 10.1016/j.patcog.2003.08.007 URL
[5]	WANG C M, WU N I, TSAI C S. A High Quality Steganographic Method with Pixel-Value Differencing and Modulus Function[J]. Journal of Systems and Software, 2008, 81(1): 150-158. doi: 10.1016/j.jss.2007.01.049 URL
[6]	LAM E Y, GOODMAN J W. A Mathematical Analysis of the DCT Coefficient Distributions for Images[J]. IEEE Transactions on Image Processing, 2000, 9(10): 1661-1666. doi: 10.1109/83.869177 pmid: 18262905
[7]	MSTAFA R J, ELLEITHY K M. Compressed and Raw Video Steganography Techniques: A Comprehensive Survey and Analysis[J]. Multimedia Tools and Applications, 2017, 76(20): 21749-21786. doi: 10.1007/s11042-016-4055-1 URL
[8]	VOULODIMOS A, DOULAMIS N, DOULAMIS A, et al. Deep Learning for Computer Vision: A Brief Review[J]. Computational Intelligence and Neuroscience, 2018, 18(1): 41-49.
[9]	OTTER D W, MEDINA J R, KALITA J K. A Survey of the Usages of Deep Learning for Natural Language Processing[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(2): 604-624. doi: 10.1109/TNNLS.5962385 URL
[10]	ZHANG Zixing, GEIGER J, POHJALAINEN J, et al. Deep Learning for Environmentally Robust Speech Recognition: An Overview of Recent Developments[J]. ACM Transactions on Intelligent Systems & Technology, 2018, 9(5): 1-28.
[11]	YANG Ziyun, WANG Zichi, ZHANG Xinpeng. A General Steganographic Framework for Neural Network Models[J]. Information Sciences, 2023, 643(9): 119-125.
[12]	YANG Zhongliang, GUO Xiaoqin, CHEN Zimin, et al. RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks[J]. IEEE Transactions on Information Forensics and Security, 2019, 14(5): 1280-1295. doi: 10.1109/TIFS.2018.2871746 URL
[13]	XU Chenyi, HUANG Lin, QIN Chuan, et al. Steganography with Constructing Neural Networks[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2025: 1-11.
[14]	WENG Xinyu, LI Yongzhi, CHI Lu. Convolutional Video Steganography with Temporal Residual Modeling[EB/OL]. (2018-02-13)[2025-05-19]. http://arxiv.org/abs/1806.02941.
[15]	VStegNET. Video Steganography Networkusing Spatio-Temporal Features and Micro-Bottleneck[EB/OL]. [2025-05-19]. https://paperswithcode.com/paper/vstegnet-video-steganography-networkusing.
[16]	JAISWAL A, KUMAR S, NIGAM A. En-VStegNET: Video Steganography Using Spatio-Temporal Feature Enhancement with 3D-CNN and Hourglass[C]// IEEE. 2020 International Joint Conference on Neural Networks (IJCNN). New York: IEEE, 2020: 1-8.
[17]	BARMAN A, DUTTA P. Facial Expression Recognition Using Reversible Neural Network[J]. Applied Soft Computing, 2024, 162: 15-26.
[18]	HAN Qinhui, JUNG C. Cross Spectral Disparity Estimation from VIS and NIR Paired Images Using Disentangled Representation and Reversible Neural Networks[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(5): 5326-5336. doi: 10.1109/TITS.2023.3238800 URL
[19]	LIU Ming, YE Guodong, ZHOU Junwei. Asymmetric Image Encryption-Hiding Scheme Based on Reversible Neural Network[J]. Science China(Technological Sciences), 2025, 68(9): 24-31.
[20]	MOU Chong, XU Youmin, SONG Jiecong, et al. Large-Capacity and Flexible Video Steganography via Invertible Neural Network[C]// IEEE. The IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2023: 22606-22615.
[21]	DINH L, KRUEGER D, BENGIO Y. NICE: Non-Linear Independent Components Estimation[EB/OL]. (2015-02-16)[2025-05-19]. http://arxiv.org/abs/1410.8516.
[22]	DINH L, SOHL-DICKSTEIN J, BENGIO S. Density Estimation Using Real NVP[EB/OL]. (2017-03-19)[2025-05-19]. http://arxiv.org/abs/1605.08803.
[23]	XU Wanni, FU Youlei, ZHU Dongmei. ResNet and Its Application to Medical Image Processing: Research Progress and Challenges[J]. Computer Methods and Programs in Biomedicine, 2023, 240: 60-76.
[24]	KOONCE B. Convolutional Neural Networks with Swift for Tensorflow: Image Recognition and Dataset Categorization[M]. Berkeley: Apress, 2021.
[25]	GIRDHAR N, SINHA A, GUPTA S. DenseNet-II: An Improved Deep Convolutional Neural Network for Melanoma Cancer Detection[J]. Soft Computing, 2023, 27(18): 13285-13304. doi: 10.1007/s00500-022-07406-z
[26]	HOU Yuntao, WU Zequan, CAI Xiaohua, et al. The Application of Improved Densenet Algorithm in Accurate Image Recognition[J]. Scientific Reports, 2024, 14(1): 24-35. doi: 10.1038/s41598-023-50497-3
[27]	WANG Xiaolong, GIRSHICK R, GUPTA A, et al. Non-Local Neural Networks[C]// IEEE. IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2018: 7794-7803.
[28]	LI Zhiyuan, WU Anhao, TAN Lin, et al. Stream Sequence Feature Fusion and Attention Mechanism Based Method for Encrypted Traffic Classification[J]. Journal of Chinese Computer Systems, 2025, 46(7): 1718-1726.
	李志远, 吴安昊, 谭林, 等. 基于流序列特征融合与注意力机制的加密流量分类方法[J]. 小型微型计算机系统, 2025, 46(7): 1718-1726.
[29]	HUANG Gao, LIU Zhuang, VAN D M L, et al. Densely Connected Convolutional Networks[C]// IEEE. IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2017: 2261-2269.
[30]	GORELICK L, BLANK M, SHECHTMAN E, et al. Actions as Space-Time Shapes[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(12): 2247-2253. pmid: 17934233
[31]	SOOMRO K, ZAMIR A R, SHAH M. UCF101: A Dataset of 101 Human Actions Classes from Videos in the Wild[EB/OL]. (2012-03-04)[2025-05-19]. http://arxiv.org/abs/1212.0402.
[32]	DAMEN D, DOUGHTY H, FARINELLA G M, et al. Rescaling Egocentric Vision: Collection, Pipeline and Challenges for EPIC-KITCHENS-100[J]. International Journal of Computer Vision, 2022, 130(1): 33-55. doi: 10.1007/s11263-021-01531-2

方法	参数/M	Cover PSNR	SSIM	Secret PSNR	SSIM
本文	0.235	73.02 /139.69	0.9999	142.02	0.9999
DRANet	16.54	31.11	0.9726	28.45	0.9293
LF-VSN	7.40	45.17	0.9800	48.39	0.9960
HCCVS	0.0387	26.77	0.9034	27.42	0.9005
VStegNET	0.0341	27.70	0.9441	23.16	0.8716
BEGAN	—	40.47	0.9794	40.66	0.9842
Liu et al	—	67.30	0.9999	69.21	0.9999

分类	小幅度干扰(PSNR/SSIM)			大幅度干扰(PSNR/SSIM)
分类	小幅度干扰(PSNR/SSIM)			0~128			0~255
分组	（1）	（2）	（3）	（1）	（2）	（3）	（1）	（2）	（3）
嵌密视频1	56.5935/ 0.9999	56.6112/ 0.9999	56.5702/ 0.9999	5.8293/ 0.0011	5.8418/ 0.0010	56.5400/ 0.9999	5.9454/ 0.0005	5.8704/ 0.0005	56.6211/ 0.9999
恢复视频1	133.9142/ 0.9999	133.9283/ 0.9999	133.9003/ 0.9999	24.4425/ 0.6423	24.5261/ 0.5180	14.4128/ 0.6399	20.7117/ 0.6271	25.5806/ 0.5502	11.5989/ : 0.5572
嵌密视频2	59.5129/ 0.9999	59.7357/ 0.9999	59.6422/ 0.9999	59.4828/ 0.9999	5.8520/ 0.0010	59.9031/ 0.9999	59.5011/ 0.9999	5.8619/ 0.0005	59.6208/ 0.9999
恢复视频2	110.1303/ 0.9999	109.9959/ 0.9999	109.9754/ 0.9999	31.5141/ 0.7704	34.3927/ 0.7793	15.9045/ 0.2545	26.4681/ 0.7535	37.2740/ 0.8634	12.3413/ 0.2274
恢复秘密	138.3023/ 0.9999	138.3258/ 0.9999	138.3489/ 0.9999	18.1464/ 0.4056	5.2470/ 0.0665	5.1961/ 0.1620	16.9346/ 0.2823	14.6085/ 0.1487	21.8954/ 0.9373
嵌入秘密	5.2372/ 0.06651	5.2269/ 0.06635	5.2133/ 0.0661	5.2357/ 0.0666	17.9618/ 0.3232	21.9229/ 0.9363	5.2681/ 0.0669	5.2508/ 0.06715	5.2018/ 0.1634