图像纹理引导的迭代水印模型

doi:10.19665/j.issn1001-2400.20240601

西安电子科技大学学报 ›› 2024, Vol. 51 ›› Issue (5): 110-121.doi: 10.19665/j.issn1001-2400.20240601

• 计算机科学与技术 & 网络空间安全 • 上一篇下一篇

图像纹理引导的迭代水印模型

武鑫婷¹^,²^,³(), 黄樱²(), 牛保宁¹(), 关虎³(), 兰方鹏¹(), 刘杰²^,³()

1.太原理工大学计算机科学与技术学院(大数据学院),山西晋中 030699
2.北京邮电大学人工智能学院,北京 100089
3.中国科学院自动化研究所,北京 100190

收稿日期:2023-11-16 出版日期:2024-07-18 发布日期:2024-07-18
通讯作者: 牛保宁(1964—),男,教授,E-mail:niubaoning@tyut.edu.cn；
黄樱(1989—),女,讲师,E-mail:ying.huang@bupt.edu.cn
作者简介:武鑫婷(1998—),女,太原理工大学博士研究生,E-mail:wuxinting1998@163.com；
关虎(1986—),男,副研究员,E-mail:hu.guan@ia.ac.cn；
兰方鹏(1979—),男,讲师,E-mail:lanfangpeng@tyut.edu.cn；
刘杰(1978—),男,副研究员,E-mail:AILJ@bupt.edu.cn
基金资助:
国家重点研发计划(2021YFF0900100);山西省重点研发计划(202302010101004);山西省基础研究计划(202203021222093);山西省基础研究计划(202203021212282);山西省重点研发计划(202102010101004);山西省研究生教育创新计划(2024KY193)

Image texture-guided iterative watermarking model

WU Xinting¹^,²^,³(), HUANG Ying²(), NIU Baoning¹(), GUAN Hu³(), LAN Fangpeng¹(), LIU Jie²^,³()

1. School of Computer Science and Technology(Big Data College),Taiyuan University of Technology,Jinzhong 030699,China
2. School of Artificial Intelligence,Beijing University of Posts and Telecommunications,Beijing 100089,China
3. Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China

Received:2023-11-16 Online:2024-07-18 Published:2024-07-18

摘要/Abstract

摘要：

近年来,深度神经网络被成功应用于数字水印领域。常见深度学习水印模型由嵌入水印的编码器、模拟攻击的噪声层、提取水印的解码器三部分构成。常用的端到端训练方式要求参与训练的攻击必须可导,此要求限制了水印模型在面对真实不可导攻击时的鲁棒性能力。此外,在一幅完整图像中,通常包含平滑纹理和粗糙纹理,然而目前的水印模型却很少直接利用其纹理信息实现水印的嵌入过程。针对上述问题,提出了一种图像纹理引导的迭代水印模型。引入图像纹理注意模块,使模型能够根据图像纹理的粗糙程度来引导水印的嵌入过程,提升水印的不可见性。采用两阶段迭代的训练方式实现对不可导攻击的学习。第一阶段,结合图像纹理注意模块,进行无攻击的端到端编-解码器联合训练,最大限度地保证水印的不可见性;第二阶段,进行不可导攻击参与的解码器独立训练,通过对任意真实攻击分布的学习,构建具有强鲁棒性的解码器。两阶段训练迭代进行,最终通过编、解码器协同训练的方式,达到模型的全局最优状态,实现水印不可见性和鲁棒性较好的平衡。实验结果表明,该模型在不可见性和鲁棒性上均优于主流的深度学习水印模型。

关键词: 深度学习, 图像水印, 纹理, 鲁棒性, 不可见性

Abstract:

Deep neural networks have been successfully used in the field of digital watermarking in recent years.An encoder that embeds the watermark,a noise layer that simulates the attacks,and a decoder that retrieves the watermark make up a typical deep learning watermarking model.The commonly used end-to-end training approach requires that the attacks involved in training should be conductible,and this requirement limits the robustness capability of watermarking models in the face of real differentiable attacks.In addition,in a complete image,which usually contains both smooth and rough textures,current watermarking models seldom utilize the texture information directly to realize the embedding process of the watermark.To address the above problem,this paper proposes an iterative watermarking model guided by the image texture.An image texture attention module is introduced so that the model can guide the embedding process of the watermark,and according to the roughness of the image texture it can improve the imperceptibility of the watermark.To make the learning of unguided attacks,a two-stage iterative training approach is adopted.In order to optimize the imperceptibility of the model,the first stage involves conducting joint end-to-end encoder-decoder training without attacks,along with the inclusion of the image texture attention module.In the second stage,the decoder involved in differentiable attacks is trained independently,and the decoder with strong robustness is built by learning from any real attack distribution.Through cooperative training of encoders and decoders,which realizes a superior balance of watermarking imperceptibility and robustness,the two-stage training is eventually carried out repeatedly to the model's global optimum.According to experimental findings,the proposed approach outperforms popular deep learning watermarking techniques in terms of both imperceptibility and robustness.

Key words: deep learning, image watermarking, texture, robustness, imperceptibility

中图分类号:

TP309.7

武鑫婷, 黄樱, 牛保宁, 关虎, 兰方鹏, 刘杰. 图像纹理引导的迭代水印模型[J]. 西安电子科技大学学报, 2024, 51(5): 110-121.

WU Xinting, HUANG Ying, NIU Baoning, GUAN Hu, LAN Fangpeng, LIU Jie. Image texture-guided iterative watermarking model[J]. Journal of Xidian University, 2024, 51(5): 110-121.

图/表 11

图1

图2

图3

图4

图5

图6

图7

表1

表2

表3

表4

参考文献 34

[1]	LEE J E, SEO Y H, KIM D W. Convolutional Neural Network-Based Digital Image Watermarking Adaptive to the Resolution of Image and Watermark[J]. Applied Sciences, 2020, 10(19):6854.
[2]	AHMADI M, NOROUZI A, KARIMI N, et al. ReDMark:Framework for Residual Diffusion Watermarking Based on Deep Networks[J]. Expert Systems with Applications, 2020, 146:113157.
[3]	LIU Y, GOU M, ZHANG J, et al. A Novel Two-Stage Separable Deep Learning Framework for Practical Blind Watermarking[C]// Proceedings of the 27th ACM International conference on multimedia. New York: ACM, 2019:1509-1517.
[4]	LUO X, ZHAN R, CHANG H, et al. Distortion Agnostic Deep Watermarking[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Piscataway:IEEE, 2020:13548-13557.
[5]	TANCIK M, MILDENHALL B, NG R. Stegastamp:Invisible Hyperlinks in Physical Photographs[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE, 2020:2117-2126.
[6]	ZHU J, KAPLAN R, JOHNSON J, et al. Hidden:Hiding Data with Deep Networks[C]// Proceedings of the European Conference on Computer Vision(ECCV). Switzerland: Springer Cham, 2018:657-672.
[7]	ZHANG H, LI Y. Digital Watermarking via Inverse Gradient Attention[C]// 2022 9th International Conference on Behavioural and Social Computing(BESC).Piscataway:IEEE, 2022:1-3.
[8]	YU C. Attention Based Data Hiding with Generative Adversarial Networks[C]// Proceedings of the AAAI Conference on Artificial Intelligence,34.Menlo Park:AAAI, 2020, 34(01):1120-1128.
[9]	HUANG J, LUO T, LI L, et al. ARWGAN:Attention-guided Robust Image Watermarking Model Based on GAN[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72:1-17.
[10]	MEI Y, WU G, YU X, et al. A Robust Blind WatermarkingScheme Based on Attention Mechanism and Neural Joint Source-Channel Coding[C]// 2022 IEEE 24th International Workshop on Multimedia Signal Processing(MMSP).Piscataway:IEEE, 2022:1-6.
[11]	FANG H, JIA Z, QIU Y, et al. De-end:Decoder-drivenWatermarking Network[J]. IEEE Transactions on Multimedia, 2023, 25:7571-7581.
[12]	JIA Z, FANG H, ZHANG W. Mbrs:Enhancing Robustness of DNN-Based Watermarking by Mini-Batch of Real and Simulated JPEG Compression[C]// Proceedings of the 29th ACM International Conference on Multimedia. New York: ACM, 2021:41-49.
[13]	WU Y, WANG B, DAI C, et al. Enhancing Robustness and Imperceptibility of Blind Watermarking with Improved Message Processor[C]//2023 IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP). Piscataway:IEEE, 2023:1-5.
[14]	MA R, GUO M, HOU Y, et al. Towards Blind Watermarking:Combining Invertible and Non-invertible Mechanisms[C]// Proceedings of the 30th ACM International Conference on Multimedia. New York: ACM, 2022:1532-1542.
[15]	ZONG T, XIANG Y, NATGUNANATHAN I, et al. Robust Histogram Shape-Based Method for Image Watermarking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2014, 25(5):717-729.
[16]	GUGELMANN D, SOMMER D, LENDERS V, et al. Screen Watermarking for Data Theft Investigationand Attribution[C]// 2018 10th International Conference on Cyber Conflict(CyCon).Piscataway:IEEE, 2018:391-408.
[17]	GHAI D, GIANEY H K, JAIN A, et al. Quantum and Dual-Tree Complex Wavelet Transform-Based Image Watermarking[J]. International Journal of Modern Physics B, 2020, 34(04):2050009.
[18]	HUANG Y, NIU B, GUAN H, et al. Enhancing Image Watermarking with Adaptive Embedding Parameter and PSNR Guarantee[J]. IEEE Transactions on Multimedia, 2019, 21(10):2447-2460.
[19]	LUO Y, LI L, LIU J, et al. A Multi-Scale Image Watermarking Based on Integer Wavelet Transform and Singular Value Decomposition[J]. Expert Systems with Applications, 2021, 168:114272.
[20]	GUL E. A Blind Robust Color Image Watermarking Method Based on Discrete Wavelet Transform and Discrete Cosine Transform Using Grayscale Watermark Image[J]. Concurrency and Computation:Practice and Experience, 2022, 34(22):e6884.
[21]	KUMAR C, SINGH A K, KUMAR P. A Recent Survey on Image Watermarking Techniques and Its Application in E-Governance[J]. Multimedia Tools and Applications, 2018, 77:3597-3622.
[22]	LAXMANIK A, SINGH P K. Robust and Imperceptible Image Watermarking Technique Based on SVD,DCT,BEMD and PSO in Wavelet Domain[J]. Multimedia Tools and Applications, 2022, 81(16):22001-22026.
[23]	WANG L, JI H. A Watermarking Optimization Method Based on Matrix Decomposition and DWT for Multi-Size Images[J]. Electronics, 2022, 11(13):2027.
[24]	POLJICAK A, DONEVSKI D, JELUSIC P B, et al. Robust DFT Watermarking Method with Gray Component Replacement Masking[J]. Multimedia Tools and Applications, 2022, 81(21):30365-30386.
[25]	PARAH S A, SHEIKH J A, LOAN N A, et al. Robust and Blind Watermarking Technique in DCT Domain Using Inter-Block Coefficient Differencing[J]. Digital Signal Processing, 2016, 53:11-24.
[26]	LU J, WANG M, DAI J, et al. Multiple Watermark Scheme Based on DWT-DCT Quantization for Medical Images[J]. Journal of Information Hiding and Multimedia Signal Processing, 2015, 6(3):458-472.
[27]	TSUI T K, ZHANG X P, ANDROUTSOS D. Color Image Watermarking Using Multidimensional Fourier Transforms[J]. IEEE Transactions on Information Forensics and Security, 2008, 3(1):16-28.
[28]	GENG L, ZHANG W, CHEN H, et al. Real-Time Attacks on Robust Watermarking Tools in the Wild by CNN[J]. Journal of Real-Time Image Processing, 2020, 17:631-641.
[29]	KANDI H, MISHRA D, GORTHI S R K S. Exploring the Learning Capabilities of Convolutional Neural Networks for Robust Image Watermarking[J]. Computers & Security, 2017, 65:247-268.
[30]	HUANG Y, GUAN H, LIU J, et al. Robust Texture-Aware Local Adaptive Image Watermarking with Perceptual Guarantee[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2023, 33(9):4660-4674.
[31]	郭娜, 黄樱, 牛保宁, 等. 融合区域边缘特征的双重自适应图像水印算法[J]. 西安电子科技大学学报, 2023, 50(5):118-131.
	GUO Na, HUANG Ying, NIU Baoning, et al. A Double Adaptive Image Watermarking Algorithm Based on Regional Edge Features[J]. Journal of Xidian University, 2023, 50(5):118-131.
[32]	黄樱, 牛保宁, 关虎, 等. 基于图像纹理的自适应水印算法[J]. 北京航空航天大学学报, 2019, 45(12):2403-2414.
	HUANG Ying, NIU Baoning, GUAN Hu, et al. Image Texture Based Adaptive Watermarking Algorithm[J]. Journal of Beijing University of Aeronautics and Astronautics, 2019, 45(12):2403-2414.
[33]	KUMAR P R, SAILAJA K L. Watermarking Algorithm Using Sobel Edge Detection[J]. International Journal of Advanced Networking and Applications, 2011, 2(05):861-867.
[34]	LIN T Y, MAIRE M, BELONGIE S, et al. Microsoft Coco:Common Objects in Context[C]// Computer Vision-ECCV 2014:13th European Conference. Switzerland: Springer Cham, 2014:740-755.

模型名称	BAR/%	PSNR/dB	SSIM
HiDDeN^[6]	74.30	35.16	0.96
HiDDeN_no	84.38	31.65	0.83
IGA^[7]	92.22	30.40	0.83
ABDH^[8]	81.24	32.80	0.86
ARWGAN^[9]	86.67	34.88	0.92
MBRS^[12]	87.64	34.33	0.89
Two-stage^[3]	88.80	33.50	0.88
ITGI	93.46	29.84	0.95

模型名称		鲁棒性(BAR)/%
模型名称		Identity		Crop p=0.035		Cropout p=0.3		Dropout p=0.3		GF σ=2.0		JPEG Q=50		Resize p=50%		CN
HiDDeN^[6]	98.10		80.34		75.96		76.89		82.72		84.09		82.72		74.30
HiDDeN_no	99.95		92.50		93.82		97.20		84.20		89.04		84.26		84.38
IGA^[7]	99.96		86.88		79.33		77.51		81.44		87.35		81.44		92.22
ABDH^[8]	95.35		81.64		74.82		75.31		79.33		82.62		80.23		81.24
ARWGAN^[9]			96.54		97.99		97.81		96.53		89.67				86.67
MBRS^[12]			92.68		99.96		99.32		95.97		91.97				87.64
Two-stage^[3]			89.00		97.30		97.40		98.60		76.20		91.70		88.80
ITGI	96.16		93.91		99.97		99.61		96.95		93.82		92.88		93.46

模型名称	是否使用迭代训练	BAR/%	SSIM
HiDDeN^[6]	No	74.30	0.96
	Yes	80.10	0.92
HiDDeN_no	No	84.38	0.83
	Yes	87.64	0.81
IGA^[7]	No	92.22	0.83
	Yes	92.53	0.80
ITGI	No	92.55	0.97
	Yes	93.46	0.95

图像纹理注意模块	BAR/%	PSNR/dB	SSIM
No	87.64	31.65	0.81
Yes	93.46	29.84	0.95

图像纹理引导的迭代水印模型

Image texture-guided iterative watermarking model

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 34

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	张静, 吴慧雪, 张少博, 李云松. 分布式策略下的解码端增强图像压缩网络[J]. 西安电子科技大学学报, 2025, 52(1): 1-13.
[2]	王潮, 蒋晓锋, 王苏敏. 面向直觉推理的量子效应交通预测算法研究[J]. 西安电子科技大学学报, 2025, 52(1): 152-162.
[3]	赵从健, 焦一源, 李雁妮. 深度语句级实体关系抽取综述[J]. 西安电子科技大学学报, 2024, 51(6): 117-131.
[4]	徐海涛, 刘玉哲, 闫欣怡, 李娇娇, 薛长斌. 一种高光谱与LiDAR特征耦合的融合分类网络[J]. 西安电子科技大学学报, 2024, 51(6): 73-83.
[5]	张铭津, 周楠, 李云松. 平滑交互式压缩网络的红外小目标检测算法[J]. 西安电子科技大学学报, 2024, 51(4): 1-14.
[6]	高迪辉, 盛立杰, 许小冬, 苗启广. 图文跨模态检索的联合特征方法[J]. 西安电子科技大学学报, 2024, 51(4): 128-138.
[7]	万鹏武, 惠茜, 陈东瑞, 吴波. 基于二维异步同相正交直方图的调制方式识别[J]. 西安电子科技大学学报, 2024, 51(4): 78-90.
[8]	管业鹏, 苏光耀, 盛怡. 双向长短期记忆网络的时间序列预测方法[J]. 西安电子科技大学学报, 2024, 51(3): 103-112.
[9]	夏译蓝, 王秀美, 程培涛. 基于多注意力机制的纹理感知视频修复方法[J]. 西安电子科技大学学报, 2024, 51(3): 136-146.
[10]	贺王鹏, 胡德顺, 李诚, 周悦, 郭宝龙. 结合模板更新与轨迹预测的孪生网络跟踪算法[J]. 西安电子科技大学学报, 2024, 51(3): 46-54.
[11]	刘伟, 王孟洋, 白宝明. 面向带宽受限场景的高效语义通信方法[J]. 西安电子科技大学学报, 2024, 51(3): 9-18.
[12]	刘振岩, 张华, 刘勇, 杨立波, 王梦迪. 一种高效的软件模糊测试种子生成方法[J]. 西安电子科技大学学报, 2024, 51(2): 126-136.
[13]	翟凤文, 孙芳林, 金静. 多尺度卷积结合Transformer的抑郁脑电分类研究[J]. 西安电子科技大学学报, 2024, 51(2): 182-195.
[14]	丁昕苗, 王家兴, 郭文. 三维注意力增强的暴力场景检测算法[J]. 西安电子科技大学学报, 2024, 51(1): 114-124.
[15]	刘博翀, 蔡怀宇, 汪毅, 陈晓冬. 用于语义分割的自监督对比式表征学习[J]. 西安电子科技大学学报, 2024, 51(1): 125-134.