Preprocessing Layer in Spatial Steganalysis Based on Deep Learning
Received date: 2018-01-25
Online published: 2018-03-31
In this paper, we propose some preprocessing methods to improve the performance of a well-designed convolution neural network based on the preprocessed layer. In the proposed methods, linear and nonlinear residuals are obtained by employing a set of derivative flters, and then quantized and truncated for the effective extraction. Experimental results show that the detection performances with the three proposed preprocessing methods are all improved. Although the improvements are not consistence under different spatial steganographic algorithms and different embedding rates. The detection performance is 6% better than the prior work for S-UNIWARD at 0.4bpp.
Key words： steganalysis; convolutional neural network; derivative flters
SHI Xiao-yu, LI Bin, TAN Shun-quan . Preprocessing Layer in Spatial Steganalysis Based on Deep Learning[J]. Journal of Applied Sciences, 2018 , 36(2) : 309 -320 . DOI: 10.3969/j.issn.0255-8297.2018.02.010
[1] Fridrich J, Kodovský J. Rich models for steganalysis of digital images[J]. IEEE Transactions on Information Forensics and Security, 2012, 7(3):868-882.
[2] Holub V, Fridrich J. Random projections of residuals for digital image steganalysis[J]. IEEE Transactions on Information Forensics and Security, 2013, 8(12):1996-2006.
[3] Tang W, Li H, Luo W, Huang J. Adaptive steganalysis against WOW embedding algorithm[C]//Proceedings of the 2nd ACM Workshop on Information Hiding and Multimedia Security, 2014:91-96.
[4] Denemark T, Sedighi V, Holub V, Cogranne R, Fridrich J. Selection-channel-aware rich model for steganalysis of digital images[C]//International Workshop on Information Forensics and Security, 2014:48-53.
[5] Li B, Wang M, Huang J, Li X. A new cost function for spatial image steganography[C]//Proceedings of the 46th IEEE Conference on Image Process, Paris, France, 2014:4206-4210.
[6] Lecun Y, Boser B, Dener J S, Henderson D, Howard R, Hubbard W, Jackel L. Handwritten digit recognition with a back propagation neural network[C]//Advances in Neural Information Processing Systems, 1990:396-404.
[7] Krizhevsky A, Sutskever I, Hinton G. Image net classifcation with deep convolutional neural networks[C]//International Conference on Neural Information Processing, 2012:1106-1114.
[8] He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition[C]//IEEE Conference on Computer Vision and Pattern Recognition, 2016:770-778.
[9] Tan S, Li B. Stacked convolutional auto-encoders for steganalysis of digital images[C]//Proceedings of Signal and Information Processing Association Annual Summit and Conference, Siem Reap, Cambodia, 2014:1-4.
[10] Qian Y, Dong J, Wang W, Tan T. Deep learning for steganalysis via convolutional neural networks[C]//Proceedings of Media Water-marking Security and Forensics, 2015:94090J.
[11] Xu G, Wu H, Shi Y. Structural design of convolutional neural networks for steganalysis[J]. IEEE Signal Processing Letters, 2016, 23(5):708-712.
[12] Ioffe S, Szegedy C. Batch normalization:accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning, 2015:448-456.
[13] Ye J, Ni J, Yi Y. Deep learning hierarchical representations for image steganalysis[J]. IEEE Transactions on Information Forensics and Security, 2017, 12(11):2545-2557.
[14] Bas P, Filler T, Pevný T. Break our steganographic system-the ins and outs of organizing BOSS[C]//Proceedings of IEEE Conference on Information Hiding, 2011:59-70.
[15] Clevert D, Unterthiner A, Hochreiter S. Fast and accurate deep network learning by exponential linear units[C]//ICLR2016, 2015, arXiv:1511.07289.
[16] Krizhevsky A, Sutskever I, Hinton G. Imagenet classifcation with deep convolutional neural networks[C]//Advances in Neural In formation Processing Systems, 2015:1097-1105.
[17] Li B, Li Z, Zhou S, Tan S, Zhang X. New steganalytic features for spatial image steganography based on derivative flters and threshold LBP operator[J]. IEEE Transactions on Information Forensics and Security, DOI:10.1109/TIFS.2017.2780805.
[18] Holub V, Fridrich J, Denemark T. Universal distortion function for steganography in an arbitrary domain[J]. Journal on Information Security, 2014(1):1.
[19] Holub V, Fridrich J. Designing steganographic distortion using directional flters[C]//Proceedings of the IEEE International Workshop on Information Forensics and Security, 2012:234-239.
[20] Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout:a simple way to prevent neural networks from overftting[J]. Journal of Machine Learning Research, 2014, 15(1):1929-1958.
/
| 〈 |
|
〉 |
