Universal Image Steganalysis Based on Convolutional Neural Network with Global Covariance Pooling
-
摘要: 1. 研究背景：图像隐写是一项可有效保护人们通信安全的技术，但易被用于不法行为。隐写分析是一项用于检测数字图像中是否含有由隐写嵌入的秘密信息的技术，可有效监控隐写术的使用。近年来，基于深度学习的隐写分析方法已经取得了比传统的基于人工设计特征的隐写分析方法更好的检测效果。
2. 目的：现有的基于深度学习的隐写分析方法大部分是基于特定域进行的，即仅针对空域或者是仅针对JPEG域的隐写分析。除此之外，已有的基于深度学习的隐写分析算法大多需要很长的训练时间来得到具有良好检测性能的隐写分析模型。为了能在隐写分析检测性能和所需训练耗时间中取得平衡，本文提出了一个可同时适用于空域和JPEG域隐写分析的高效隐写分析网络。
3. 方法：我们对隐写分析网络中会对检测性能造成影响的部分进行了精心地设计，如在预处理模块中为提高输入图像的信噪比，引入了SRM和Gabor高通滤波器；在特征提取模块中的卷积层组，引入了有助于梯度传播和特征重用的跳跃连接；在输出特征进入分类模块前，采用了全局协方差池化来进一步聚合高层次特征，使所提取特征的表达能力更强。我们在常用的BOSSBase 和BOWS2数据集上进行相应的实验验证。
4. 结果：大量的实验结果证明，所提隐写分析网络在空域上的检测性能已超越目前一些先进的隐写分析方法，如YeNet、ZhuNet和SRNet，在频域上的检测性能也与相应的隐写分析方法如J-YeNet、SRNet相当，但所需的训练耗时却大大减少了，约为YeNet、J-YeNet、ZhuNet的1/2，SRNet的1/4。
5. 结论：我们提出了一个简单且高效的隐写分析网络，该网络在空域上的隐写分析检测性能已超越目前先进的隐写分析网络，在频域上的检测性能也与目前国际领先水平相当，且所需要的训练耗时大大减少了。在进一步的工作中，我们将会考虑如何将通道选择信息融入所提网络，以及如何减少所提网络中启发式的设计。Abstract: Recently, steganalytic methods based on deep learning have achieved much better performance than traditional methods based on handcrafted features. However, most existing methods based on deep learning are specially designed for one image domain (i.e., spatial or JPEG), and they often take long time to train. To make a balance between the detection performance and the training time, in this paper, we propose an effective and relatively fast steganalytic network called US-CovNet (Universal Steganalytic Covariance Network) for both {the} spatial and JPEG domains. To this end, we carefully design several important components of {US-CovNet} that will significantly affect the detection performance, including the high-pass filter set, the shortcut connection and the pooling {layer}. Extensive experimental results show that compared with the current best steganalytic networks (i.e., SRNet and J-YeNet), {US-CovNet} can achieve the state-of-the-art results for detecting spatial steganography and have competitive performance for detecting JPEG steganography. For example, the detection accuracy of US-CovNet is at least 0.56% higher than that of SRNet in the spatial domain. In the JPEG domain, US-CovNet performs slightly worse than J-YeNet in some cases with the degradation of less than 0.78%. However, the training time of US-CovNet is significantly reduced, which is less than 1/4 and 1/2 of SRNet and J-YeNet respectively. -
-
[1] 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.
[2] Kodovsky J, Fridrich J, Holub V. Ensemble classifiers for steganalysis of digital media. IEEE Transactions on Information Forensics and Security, 2012, 7(2): 432-444. DOI: 10.1109/TIFS.2011.2175919.
[3] Holub V, Fridrich J. Random projections of residuals for digital image steganalysis. IEEE Transactions on Information Forensics and Security, 2013, 8(12): 1996-2006. DOI: 10.1109/TIFS.2013.2286682.
[4] Denemark T, Sedighi V, Holub V, Cogranne R, Fridrich J. Selection-channel-aware rich model for steganalysis of digital images. In Proc. the IEEE International Workshop on Information Forensics and Security, December 2014, pp.48-53. DOI: 10.1109/WIFS.2014.7084302.
[5] Song X, Liu F, Yang C, Luo X, Zhang Y. Steganalysis of adaptive JPEG steganography using 2D Gabor filters. In Proc. the 3rd ACM Workshop on Information Hiding and Multimedia Security, June 2015, pp.15-23. DOI: 10.1145/2756601.2756608.
[6] Denemark T, Boroumand M, Fridrich J. Steganalysis features for content-adaptive JPEG steganography. IEEE Transactions on Information Forensics and Security, 2016, 11(8): 1736-1746. DOI: 10.1109/TIFS.2016.2555281.
[7] Tan S, Li B. Stacked convolutional auto-encoders for steganalysis of digital images. In Proc. the Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, December 2014. DOI: 10.1109/APSIPA.2014.7041565.
[8] Qian Y, Dong J, Wang W, Tan T. Deep learning for steganalysis via convolutional neural networks. In Proc. the 2015 SPIE/IS&T Electronic Imaging Symposium, February 2015, Article No. 94090J. DOI: 10.1117/12.2083479.
[9] Xu G, 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.
[10] Ye J, Ni J, 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.
[11] Huang J, Ni J, Wan L, Yan J. A customized convolutional neural network with low model complexity for JPEG steganalysis. In Proc. the ACM Workshop on Information Hiding and Multimedia Security, July 2019, pp.198-203. DOI: 10.1145/3335203.3335734.
[12] Boroumand M, Chen M, Fridrich J. Deep residual network for steganalysis of digital images. IEEE Transactions on Information Forensics and Security, 2019, 14(5): 1181-1193. DOI: 10.1109/TIFS.2018.2871749.
[13] Yedroudj M, Comby F, Chaumont M. Yedroudj-Net: An efficient CNN for spatial steganalysis. In Proc. the IEEE International Conference on Acoustics, Speech and Signal Processing, April 2018, pp.2092-2096. DOI: 10.1109/ICASSP.2018.8461438.
[14] Ye D, Jiang S, Li S, Liu C. Faster and transferable deep learning steganalysis on GPU. Journal of Real-Time Image Processing, 2019, 16(3): 623-633. DOI: 10.1007/s11554-019-00870-1.
[15] Li S, Ye D, Jiang S, Liu C, Niu X, Luo X. Anti-steganalysis for image on convolutional neural networks. Multimedia Tools and Applications, 2020, 79(7): 4315-4331. DOI: 10.1007/s11042-018-7046-6.
[16] Jiang S, Ye D, Huang J, Shang Y, Zheng Z. SmartSteganogaphy: Light-weight generative audio steganography model for smart embedding application. Journal of Networking and Computer Applications, 2020, 165: Article No. 102689. DOI: 10.1016/j.jnca.2020.102689.
[17] Kodovský J, Fridrich J, Holub V. On dangers of overtraining steganography to incomplete cover model. In Proc. the 13th ACM Multimedia Workshop on Multimedia and Security, September 2011, pp.69-76. DOI: 10.1145/2037252.2037266.
[18] Zhang R, Zhu F, Liu J, Liu G. Depth-wise separable convolutions and multi-level pooling for an efficient spatial CNN-based steganalysis. IEEE Transactions on Information Forensics and Security, 2020, 15: 1138-1150. DOI: 10.1109/TIFS.2019.2936913.
[19] Li Q, Feng G, Wu H, Zhang X. Ensemble steganalysis based on deep residual network. In Proc. the 18th International Workshop on Digital-Forensics and Watermarking, November 2019, pp.84-95. DOI: 10.1007/978-3-030-43575-2.
[20] Su A, Zhao X. Boosting image steganalysis under universal deep learning architecture incorporating ensemble classification strategy. IEEE Signal Processing Letters, 2019, 26(12): 1852-1856. DOI: 10.1109/LSP.2019.2950081.
[21] Kim J, Park H, Park J. CNN-based image steganalysis using additional data embedding. Multimedia Tools and Applications, 2020, 79(1/2): 1355-1372. DOI: 10.1007/s11042-019-08251-3.
[22] Li Q, Shao Z, Tan S, Zeng J, Li B. Non-structured pruning for deep-learning based steganalytic frameworks. In Proc. the Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, November 2019, pp.1735-1739. DOI: 10.1109/APSIPAASC47483.2019.9023018.
[23] Deng X, Chen B, Luo W, Luo D. Fast and effective global covariance pooling network for image steganalysis. In Proc. the ACM Workshop on Information Hiding and Multimedia Security, July 2019, pp.230-234. DOI: 10.1145/3335203.3335739.
[24] Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proc. the 32nd International Conference on Machine Learning, July 2015, pp.448-456.
[25] Li P, Xie J, Wang Q, Gao Z. Towards faster training of global covariance pooling networks by iterative matrix square root normalization. In Proc. the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 2018, pp.947-955. DOI: 10.1109/CVPR.2018.00105.
[26] He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In Proc. the IEEE Conference on Computer Vision and Pattern Recognition, June 2016, pp.770-778. DOI: 10.1109/CVPR.2016.90.
[27] Wu S, Zhong S, Liu Y. Steganalysis via deep residual network. In Proc. the 22nd IEEE International Conference on Parallel and Distributed Systems, December 2016, pp.1233-1236. DOI: 10.1109/ICPADS.2016.0167.
[28] Li P, Xie J, Wang Q, Zuo W. Is second-order information helpful for large-scale visual recognition? In Proc. the IEEE International Conference on Computer Vision, October 2017, pp.2089-2097. DOI: 10.1109/ICCV.2017.228.
[29] Bas P, Filler T, Pevný T. “Break our steganographic system”: The ins and outs of organizing BOSS. In Proc. the 13th International Workshop on Information Hiding, May 2011, pp.59-70. DOI: 10.1007/978-3-642-24178-9.
[30] He K, Zhang X, Ren S, Sun J. Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification. In Proc. the IEEE International Conference on Computer Vision, December 2015, pp.1026-1034. DOI: 10.1109/ICCV.2015.123.
[31] Bengio Y, Louradour J, Collobert R, Weston J. Curriculum learning. In Proc. the 26th International Conference on Machine Learning, June 2009, pp.41-48. DOI: 10.1145/1553374.1553380.
[32] Holub V, Fridrich J, Denemark T. Universal distortion function for steganography in an arbitrary domain. EURASIP Journal on Information Security, 2014, 2014: Article No. 1. DOI: 10.1186/1687-417X-2014-1.
[33] Li B, Wang M, Huang J, Li X. A new cost function for spatial image steganography. In Proc. the IEEE International Conference on Image Processing, October 2014, pp.4206-4210. DOI: 10.1109/ICIP.2014.7025854.
[34] Holub V, Fridrich J. Designing steganographic distortion using directional filters. In Proc. the IEEE International Workshop on Information Forensics and Security, December 2012, pp.234-239. DOI: 10.1109/WIFS.2012.6412655.
[35] Guo L, Ni J, Shi Y Q. Uniform embedding for efficient JPEG steganography. IEEE Transactions on Information Forensics and Security, 2014, 9(5): 814-825. DOI: 10.1109/TIFS.2014.2312817.
[36] Denemark T, Fridrich J, Comesaña-Alfaro P. Improving selection-channel-aware steganalysis features. Electronic Imaging, 2016, 28: Article No. art00016. DOI: 10.2352/ISSN.2470-1173.2016.8.MWSF-080.
[37] Elsken T, Metzen J H, Hutter F. Neural architecture search: A survey. Journal of Machine Learning Research, 2019, 20: Article No. 55.
-
期刊类型引用(13)
1. G. Vijitha, B. Sargunam. AN APPROACH FOR IMAGE STEGANOGRAPHY AND STEGANALYSIS USING REGRESSIVE STUDENT PSYCHOLOGY OPTIMIZATION-ENABLED DEEP MAX-OUT NETWORK. Biomedical Engineering: Applications, Basis and Communications, 2025, 37(02) 
必应学术
2. Jian He, Shaowei Weng, Lifang Yu, et al. Steganalysis Network With Two-Branch Preprocessing for Spatial and JPEG Domains. IEEE Transactions on Circuits and Systems for Video Technology, 2025, 35(2): 1451. 必应学术
3. Tong Fu, Liquan Chen, Yinghua Jiang, et al. Image Steganalysis Based on Dual-Path Enhancement and Fractal Downsampling. IEEE Transactions on Information Forensics and Security, 2025, 20: 1. 必应学术
4. Kang-Kang Wei, Wei-Qi Luo, Shun-Quan Tan, et al. CTNet: A Convolutional Transformer Network for Color Image Steganalysis. Journal of Computer Science and Technology, 2025, 40(2): 413. 必应学术
5. Ntivuguruzwa Jean de la Croix, Tohari Ahmad, Fengling Han, et al. HSDetect-Net: A Fuzzy-Based Deep Learning Steganalysis Framework to Detect Possible Hidden Data in Digital Images. IEEE Access, 2025, 13: 43013. 必应学术
6. Tong Fu, Liquan Chen, Yuan Gao, et al. DCANet: CNN model with dual-path network and improved coordinate attention for JPEG steganalysis. Multimedia Systems, 2024, 30(4) 必应学术
7. Zhiqiang Wu, Shuhui Wan. An Image Steganalysis Algorithm Based on Multi-Resolution Feature Fusion. International Journal of Information Security and Privacy, 2024, 18(1): 1. 必应学术
8. Numrena Farooq, Roohie Naaz Mir. Image Steganalysis using Deep Convolution Neural Networks: A Literature Survey. International Journal of Sensors, Wireless Communications and Control, 2024, 14(4): 247. 必应学术
9. Hamza Kheddar, Mustapha Hemis, Yassine Himeur, et al. Deep learning for steganalysis of diverse data types: A review of methods, taxonomy, challenges and future directions. Neurocomputing, 2024, 581: 127528. 必应学术
10. Ntivuguruzwa Jean de La Croix, Tohari Ahmad. Toward Hidden Data Detection via Local Features Optimization in Spatial Domain Images. 2023 Conference on Information Communications Technology and Society (ICTAS), 必应学术
11. Akram Kadhim Mshet, Huda Kadhim Tayyeh. Deep learning-based image Steganalysis. CYBER DEFENSE IRAQ: CDI2024, 必应学术
12. Mehrab Hosain, Rajiv Kapoor. Advances in Artificial-Business Analytics and Quantum Machine Learning. Lecture Notes in Electrical Engineering, 必应学术
13. Yuanlin Yang. Large Capacity Data Hiding in Binary Image black and white mixed regions. 2023 3rd International Conference on Electronic Information Engineering and Computer (EIECT), 必应学术
其他类型引用(0)
-
其他相关附件
-
本文英文pdf
2022-5-9-0572-Highlights 点击下载(657KB) -
本文附件外链
https://rdcu.be/cX761
-
计量
- 文章访问数: 114
- HTML全文浏览量: 3
- PDF下载量: 2
- 被引次数: 13
下载:
