基于深度强化学习的自适应图像隐写算法

doi:10.19678/j.issn.1000-3428.0068359

摘要/Abstract

摘要：

针对自适应图像隐写中如何在轻量化隐写、最佳嵌入定位、高隐匿输出三者之间实现均衡的问题, 提出一种基于深度强化学习的自适应图像隐写算法(AISA-DRL)。设计一种轻量化安全隐写网络, 在降低模型隐写成本的前提下加强模型对图像隐写特征的提取能力, 增强载密图像的安全性和稳定性。首先将具有高效特征融合特性的EPSANet引入EfficientnetV2-s, 得到改进的EPSA-EfficientnetV2-s, 以提高像素级嵌入过程的表征能力, 从而获得最优像素修改位张量。随后将秘密信息与最优像素修改位张量加权求和得到载密图像。最后通过学习隐写分析网络对载密图像进行最优像素级奖励分配, 根据设计的最小化失真函数通过梯度回传来更新网络参数, 以获得最佳嵌入位置, 从而实现秘密信息的最佳化嵌入。实验结果表明, AISA-DRL算法的模型参数量减少了94.22%, FLOPs减少了24.88%, 与其他基于强化学习的隐写方案相比, 在不同经典隐写分析器下的检错率提高了2.48%~6.55%。此外, 在不同载荷下生成的载密图像PSNR值均在30 dB以上, 不仅提高了模型对像素修改位的定位准确率, 而且使隐写网络具有更强的表征能力。

关键词: 图像隐写, 强化学习, 隐写分析, 嵌入策略, 特征提取

Abstract:

This study proposes an Adaptive Image Steganography Algorithm based on Deep Reinforcement Learning (AISA-DRL) to realize a balance among lightweight steganography, optimal embedding positioning, and high hidden outputs. A lightweight secure steganography network is designed to enhance the ability of the model to extract stego features of an image based on the premise of reducing the steganography cost and optimizing the security and stability of the stego image. First, an EPSANet with efficient feature fusion characteristics is introduced into EfficientnetV2-s to obtain an improved EPSA-EfficientnetV2-s. This integration enhances the representation ability of the pixel-level embedding process to obtain the optimal tensor for the pixel modification bit. The stego image is then calculated using the weighted sum of the secret information and the optimal tensor of the pixel modification. Finally, by learning the steganographic analysis network, the optimal pixel-level reward is assigned to the stego image, and the network parameters are updated by gradient backpropagation according to the designed minimum distortion function. This process ensures that the best embedding position is obtained, and the optimal embedding of the secret information is realized. The experimental results show that the parameters of the AISA-DRL are reduced by 94.22%, and FLOPs are reduced by 24.88%. Compared with steganographic methods based on reinforcement learning, the error detection rate of the AISA-DRL increases by 2.48%-6.55%. Additionally, the PSNR values of the stego images generated under different loads are all greater than 30 dB. Thus, the proposed AISA-DRL improves the locational accuracy of the model for the pixel modification bit. The steganographic network of the AISA-DRL demonstrates a stronger characterization capability compared to other algorithms.

Key words: image steganography, reinforcement learning, steganography analysis, embedding strategy, feature extraction

钱清, 龙永, 蒋忠远, 段春红, 王宏. 基于深度强化学习的自适应图像隐写算法[J]. 计算机工程, 2024, 50(8): 319-327.

Qing QIAN, Yong LONG, Zhongyuan JIANG, Chunhong DUAN, Hong WANG. Adaptive Image Steganography Algorithm Based on Deep Reinforcement Learning[J]. Computer Engineering, 2024, 50(8): 319-327.

https://www.ecice06.com/CN/Y2024/V50/I8/319

图/表 9

图1 AISA-DRL模型的整体框架结构

Fig.1 Overall framework structure of AISA-DRL model

图2 不同主干网络下的隐写分析检测错误率

Fig.2 Steganalysis detection error rate under different backbone networks

图3 检测错误率对比结果

Fig.3 Comparison results of detection error rate

图4 AISA-DRL隐写实际效果

Fig.4 AISA-DRL steganography actual effect

图5 载密图像隐写质量分析结果

Fig.5 Analysis results of steganographic quality of stego images

参考文献 31

1	VEERA MANINDRA A P, KARTHIKEYAN B. OTP camouflaging using LSB steganography and public key cryptography[C]//Proceedings of the 3rd International Conference on Electronics and Sustainable Communication Systems. Washington D. C., USA: IEEE Press, 2022: 109-115.
2	HASSABALLAH M, ABDEL HAMEED M, AWAD A I, et al. A novel image steganography method for industrial Internet of things security. IEEE Transactions on Industrial Informatics, 2021, 17(11): 7743- 7751. doi: 10.1109/TII.2021.3053595
3	MOHSIN A H, ZAIDAN A A, ZAIDAN B B, et al. PSO: blockchain-based image steganography: towards a new method to secure updating and sharing COVID-19 data in decentralised hospitals intelligence architecture. Multimedia Tools and Applications, 2021, 80(9): 14137- 14161. doi: 10.1007/s11042-020-10284-y
4	RAHMAN S, MASOOD F, ULLAH KHAN W, et al. A novel approach of image steganography for secure communication based on LSB substitution technique. Computers, Materials & Continua, 2020, 64(1): 31- 61.
5	VOLETI L, BALAJEE R M, VALLEPU S K, et al. A secure image steganography using improved LSB technique and Vigenere cipher algorithm[C]//Proceedings of International Conference on Artificial Intelligence and Smart Systems. Washington D. C., USA: IEEE Press, 2021: 1005-1010.
6	KUMAR M, SONI A, SHEKHAWAT A R S, et al. Enhanced digital image and text data security using hybrid model of LSB steganography and AES cryptography technique[C]//Proceedings of the 2nd International Conference on Artificial Intelligence and Smart Energy. Washington D. C., USA: IEEE Press, 2022: 1453-1457.
7	ALI H A, JALIL A J, HUSSEIN M K. Vernam encryption and steganography of a number of images in the digital video[C]//Proceedings of International Conference on Electrical, Computer, Communications and Mechatronics Engineering. Washington D. C., USA: IEEE Press, 2021: 1-6.
8	QIN X H, LI B, HUANG J W. A new spatial steganographic scheme by modeling image residuals with multivariate Gaussian model[C]//Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing. Washington D. C., USA: IEEE Press, 2019: 2617-2621.
9	余恒, 王让定, 严迪群, 等. 基于采样值排序的音频可逆隐写算法. 计算机工程, 2021, 47(1): 123-128, 138. URL
	YU H, WANG R D, YAN D Q, et al. Reversible audio steganography algorithm based on sample value ordering. Computer Engineering, 2021, 47(1): 123-128, 138. URL
10	张祎, 罗向阳, 王金伟, 等. 数字图像鲁棒隐写综述. 中国图象图形学报, 2022, 27(1): 3- 26. URL
	ZHANG Y, LUO X Y, WANG J W, et al. Research progress on digital image robust steganography. Journal of Image and Graphics, 2022, 27(1): 3- 26. URL
11	刘帅威, 李智, 王国美, 等. 基于Transformer和GAN的对抗样本生成算法. 计算机工程, 2024, 50(2): 180- 187. URL
	LIU S W, LI Z, WANG G M, et al. Adversarial example generation algorithm based on transformer and GAN. Computer Engineering, 2024, 50(2): 180- 187. URL
12	高培贤, 魏立线, 刘佳, 等. 针对图像隐写分析的卷积神经网络结构改进. 计算机工程, 2018, 44(10): 309- 313. URL
	GAO P X, WEI L X, LIU J, et al. Improvement of convolutional neural network structure for image steganalysis. Computer Engineering, 2018, 44(10): 309- 313. URL
13	TANG W X, TAN S Q, LI B, et al. Automatic steganographic distortion learning using a generative adversarial network. IEEE Signal Processing Letters, 2017, 24(10): 1547- 1551. doi: 10.1109/LSP.2017.2745572
14	YANG J H, RUAN D Y, HUANG J W, et al. An embedding cost learning framework using GAN. IEEE Transactions on Information Forensics and Security, 2020, 15, 839- 851. doi: 10.1109/TIFS.2019.2922229
15	TANG W X, LI B, BARNI M, et al. An automatic cost learning framework for image steganography using deep reinforcement learning. IEEE Transactions on Information Forensics and Security, 2021, 16, 952- 967. doi: 10.1109/TIFS.2020.3025438
16	XU G S, WU H Z, SHI Y Q. Structural design of convolutional neural networks for steganalysis. IEEE Signal Processing Letters, 2016, 23(5): 708- 712. doi: 10.1109/LSP.2016.2548421
17	MO X B, TAN S Q, LI B, et al. MCTSteg: a Monte Carlo tree search-based reinforcement learning framework for universal non-additive steganography. IEEE Transactions on Information Forensics and Security, 2021, 16, 4306- 4320. doi: 10.1109/TIFS.2021.3104140
18	ZHANG H, ZU K K, LU J, et al. EPSANet: an efficient pyramid squeeze attention block on convolutional neural network[M]. Berlin, Germany: Springer, 2023.
19	BAS P, FILLER T, PEVNY T. Break our steganographic system: the ins and outs of organizing BOSS[M]. Berlin, Germany: Springer, 2011.
20	WANG Z C, FENG G R, SHEN L Q, et al. Cover selection for steganography using image similarity. IEEE Transactions on Dependable and Secure Computing, 2022, 20(3): 2328- 2340.
21	DENG J, DONG W, SOCHER R, et al. ImageNet: a large-scale hierarchical image database[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2009: 248-255.
22	DUAN X, LIU N, GOU M, et al. SteganoCNN: image steganography with generalization ability based on convolutional neural network. Entropy, 2020, 22(10): e1140. doi: 10.3390/e22101140
23	FRIDRICH J, KODOVSKY J. Rich models for steganalysis of digital images. IEEE Transactions on Information Forensics and Security, 2012, 7(3): 868- 882. doi: 10.1109/TIFS.2012.2190402
24	DENEMARK T, SEDIGHI V, HOLUB V, et al. Selection-channel-aware rich model for steganalysis of digital images[C]//Proceedings of IEEE International Workshop on Information Forensics and Security. Washington D. C., USA: IEEE Press, 2014: 48-53.
25	YE J, NI J Q, YI Y. Deep learning hierarchical representations for image steganalysis. IEEE Transactions on Information Forensics and Security, 2017, 12(11): 2545- 2557. doi: 10.1109/TIFS.2017.2710946
26	WANG Z, FENG G, QIAN Z, et al. JPEG steganography with content similarity evaluation. IEEE Transactions on Cybernetics, 2023, 53(8): 5082- 5093. doi: 10.1109/TCYB.2022.3155732
27	QIAO T, LUO X Y, WU T, et al. Adaptive steganalysis based on statistical model of quantized DCT coefficients for JPEG images. IEEE Transactions on Dependable and Secure Computing, 2021, 18(6): 2736- 2751. doi: 10.1109/TDSC.2019.2962672
28	TANG W X, LI B, LI W X, et al. Reinforcement learning of non-additive joint steganographic embedding costs with attention mechanism. Science China Information Sciences, 2023, 66(3): 132305. doi: 10.1007/s11432-021-3453-5
29	SONG T T, LIU M L, LUO W Q, et al. Enhancing image steganography via stego generation and selection[C]//Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing. Washington D. C., USA: IEEE Press, 2021: 2695-2699.
30	LUO J, HE P S, LIU J Y, et al. Content-adaptive adversarial embedding for image steganography using deep reinforcement learning[C]//Proceedings of IEEE International Conference on Multimedia and Expo. Washington D. C., USA: IEEE Press, 2023: 49-54.
31	LI B, WANG M, HUANG J W, et al. A new cost function for spatial image steganography[C]//Proceedings of IEEE International Conference on Image Processing. Washington D. C., USA: IEEE Press, 2014: 4206-4210.

[1]	赵俊涛, 李陶深, 卢志翔. 基于最优近邻的局部保持投影方法[J]. 计算机工程, 2024, 50(9): 161-168.
[2]	石琼, 段辉, 师智斌. 基于深度强化学习的可信任务卸载方案[J]. 计算机工程, 2024, 50(8): 142-152.
[3]	胡庆. 多尺度融合与双输出U-Net网络的行人重识别[J]. 计算机工程, 2024, 50(6): 102-109.
[4]	高家豪, 胡创业, 丁男, 刘战东. 智能网联汽车中联合驾驶风格的交通流数据有效性分析[J]. 计算机工程, 2024, 50(6): 367-376.
[5]	孙文洁, 李宗民, 孙浩淼. 基于图神经网络的多智能体强化学习值函数分解方法[J]. 计算机工程, 2024, 50(5): 62-70.
[6]	傅明建, 郭福强. 基于深度强化学习的无信号灯路口决策研究[J]. 计算机工程, 2024, 50(5): 91-99.
[7]	梁松林, 林伟, 王珏, 杨庆. 面向后渗透攻击行为的网络恶意流量检测研究[J]. 计算机工程, 2024, 50(5): 128-138.
[8]	张斯力, 李梓健, 蔡瑞初, 郝志峰, 闫玉光. 基于因果机制约束的强化推荐系统[J]. 计算机工程, 2024, 50(5): 279-290.
[9]	李振鲁, 黄威, 孙锴. 复杂环境下的轻量化道路目标识别算法研究[J]. 计算机工程, 2024, 50(4): 219-227.
[10]	袁文涛, 卫文韬, 高德民. 融合注意力机制的多视图卷积手势识别研究[J]. 计算机工程, 2024, 50(3): 208-215.
[11]	冯雄波, 黄于欣, 赖华, 高玉梦. 基于多策略强化学习的低资源跨语言摘要方法研究[J]. 计算机工程, 2024, 50(2): 68-77.
[12]	杜海军, 余粟. 基于时空图注意力网络的服务机器人动态避障[J]. 计算机工程, 2024, 50(2): 105-112.
[13]	任义, 苏博, 袁帅. 教育领域下多维度特征命名实体识别方法[J]. 计算机工程, 2024, 50(10): 110-118.
[14]	张俊娜, 韩超臣, 陈家伟, 赵晓焱, 袁培燕. 一种联合边缘服务器部署与服务放置的方法[J]. 计算机工程, 2024, 50(10): 266-280.
[15]	孔月萍, 杨世海, 段梅梅, 丁泽诚, 方凯杰. 基于混合动作强化学习的电动汽车聚合商决策优化算法[J]. 计算机工程, 2024, 50(10): 418-428.

选择文件类型/文献管理软件名称

选择包含的内容