Image stegoanalysis using deep neural networks and~heteroassociative integral transformations

M. A. Dryuchenko ; A. A. Sirota

Prikladnaâ diskretnaâ matematika, no. 1 (2022), pp. 35-58.

Résumé

The problem of steganalysis of digital images is considered. The proposed approach is based on the use of deep convolutional neural networks with a relatively simple architecture, distinguished by the use of additional layers of special processing. These networks are trained and used for steganalysis of small fragments of the original large images. For the analysis of full sized images, it is proposed to carry out secondary post-processing, which involves combining the obtained classification results in blocks as a sequence of binary features according to the scheme of a naive Bayesian classifier. We propose to use integral heteroassociative transformations that provide the selection of the estimated and stochastic (masking) components on the processed image fragment based on the forecast model of one part of the fragment in relation to another to identify violations of the structural and statistical image properties after message embedding. Such transformations are included in the architecture of trained neural networks as an additional layer. Alternative versions of deep neural network architectures (with and without an integral layer of heteroassociative transformation) are considered. The PPG-LIRMM-COLOR images base was used to create data sets. Experiments have been carried out for several well-known stego algorithms (including the classic block and block-spectral algorithms of Kutter, Koha — Zhao, modern algorithms EMD, MBEP and algorithms for adaptive spatial steganography WOW and S-UNIWARD) and for the stego algorithms based on the use of heteroassociative compression transformations. It is shown that the accuracy of steganalysis obtained when implementing the proposed information processing schemes for large images with relatively low computational costs is comparable to the results obtained by other authors, and in some cases even exceeds them.

Détail

Keywords: steganography, steganalysis, machine learning, deep neural networks.

Bibliographie

[1] O. I. Shelukhin, Steganografiya. Algoritmy i programmnaya realizatsiya, Goryachaya liniya Telekom, M., 2017, 592 pp.

[2] B. Czaplewski, “Current trends in the field of steganalysis and guidelines for constructions of new steganalysis schemes”, Przegla̧d Telekomunikacyjny + Wiadomósci Telekomunikacyjne, 2017, no. 10, 1121–1125

[3] S. Lyu, H. Farid, “Detecting hidden messages using higher-order statistics and support vector machines”, Intern. Workshop Inform. Hiding, Springer, Berlin–Heidelberg, 2002, 340–354 https://farid.berkeley.edu/downloads/publications/ih02.pdf

[4] S. Lyu, H. Farid, “Steganalysis using color wavelet statistics and one-class support vector machines” (California, USA, 2004), Proc. SPIE, 2004, 35–45 https://www.researchgate.net/publication/221011180_Steganalysis_using_color_wavelet_statistics_and_one-class_support_vector_machines | DOI

[5] T. Pevny, P. Bas, J. Fridrich, “Steganalysis by subtractive pixel adjacency matrix”, IEEE Trans. Inform. Forensics Security, 5:2 (2010), 215–224 | DOI

[6] J. Fridrich, “Rich models for steganalysis of digital images”, IEEE Trans. Inform. Forensics Security, 7:3 (2012), 868–882 | DOI

[7] V. Holub, J. Fridrich, “Random projections of residuals for digital image steganalysis”, IEEE Trans. Inform. Forensics Security, 8:12 (2013), 1996–2006 | DOI

[8] P. Bas, T. Filler, T. Pevny, “Break our steganographic system the ins and outs of organizing BOSS”, LNCS, 6958, 2011, 59–70

[9] PPG-LIRMM-COLOR base, http://www.lirmm.fr/chaumont/PPG-LIRMM-COLOR.html

[10] T. Pevny, T. Filler, P. Bas, “Using high-dimensional image models to perform highly undetectable steganography”, LNCS, 6387, 2010, 161–177

[11] V. Holub, J. Fridrich, “Digital image steganography using universal distortion”, Proc. 1st ACM Workshop IHMMSec, ACM, 2013, 59–68

[12] V. Holub, J. Fridrich, “Designing steganographic distortion using directional filters”, Proc. 4th IEEE Intern. Workshop WIFS, 2012, 234–239

[13] J. Kodovsky, J. Fridrich, V. Holub, “Ensemble classifiers for steganalysis of digital media”, IEEE Trans. Inform. Forensics Security, 7:2 (2010), 434–444 | MR

[14] V. A. Monarev, A. I. Pestunov, “Effektivnoe obnaruzhenie steganograficheski skrytoi informatsii posredstvom integralnogo klassifikatora na osnove szhatiya dannykh”, Prikladnaya diskretnaya matematika, 2018, no. 40, 59–71 | MR

[15] R. Tabares-Soto, R. Ramos-Pollan, G. Isaza, “Deep learning applied to steganalysis of digital images: A systematic review”, IEEE Access, 7 (2019), 68970–68990 | DOI

[16] Y. Qian, J. Dong, W. Wang, T. Tan, “Deep learning for steganalysis via convolutional neural networks”, Proc. Int. Symp. Electron. Imag., 9409 (2015), 94090J

[17] M. Yedroudj, F. Comby, M. Chaumont, “Yedrouj-Net: An efficient CNN for spatial steganalysis”, Proc. IEEE Intern. Conf. Acoustics, Speech Signal Processing, 2018, 2092–2096

[18] R. Zhang, F. Zhu, J. Liu, G. Liu, Efficient Feature Learning and Multisizeimage Steganalysis Based on CNN, 2018, arXiv: 1807.11428

[19] A. A. Polunin, E. A. Yandashevskaya, “Ispolzovanie apparata svertochnykh neironnykh setei dlya stegoanaliza tsifrovykh izobrazhenii”, Trudy ISP RAN, 32:4 (2020), 155–164

[20] A. A. Sirota, M. A. Dryuchenko, “Obobschennye algoritmy szhatiya izobrazhenii na fragmentakh proizvolnoi formy i ikh realizatsiya s ispolzovaniem iskusstvennykh neironnykh setei”, Kompyuternaya optika, 2015, no. 5, 751–761

[21] M. A. Dryuchenko, A. A. Sirota, “Interpolation and masking effects of heteroassociative compressive transformations”, J. Phys.: Conf. Ser., 1902 (2020), 1–10 | DOI

[22] M. A. Dryuchenko, A. A. Sirota, “Geteroassotsiativnye szhimayuschie preobrazovaniya tsifrovykh izobrazhenii i ikh interpoliruyuschie i maskiruyuschie svoistva”, Sb. trudov Mezhdunar. nauch.-tekhn. konf. «Aktualnye problemy prikladnoi matematiki, informatiki i mekhaniki» (Voronezh, 07–09 dekabrya 2020), 312–322

[23] A. A. Sirota, M. A. Dryuchenko, E. Yu. Mitrofanova, “Metod sozdaniya tsifrovykh vodyanykh znakov na osnove geteroassotsiativnykh szhimayuschikh preobrazovanii izobrazhenii i ego realizatsiya s ispolzovaniem iskusstvennykh neironnykh setei”, Kompyuternaya optika, 2018, no. 3, 483–494

[24] A. A. Sirota, M. A. Dryuchenko, E. Yu. Mitrofanova, “Neirosetevye funktsionalnye modeli i algoritmy preobrazovaniya informatsii dlya sozdaniya tsifrovykh vodyanykh znakov”, Izv. vuzov. Radioelektronika, 2015, no. 1, 3–16

[25] A. A. Sirota, M. A. Dryuchenko, A. Yu. Ivankov, “Stegoanaliz tsifrovykh izobrazhenii s ispolzovaniem metodov poverkhnostnogo i glubokogo mashinnogo obucheniya: izvestnye podkhody i novye resheniya”, Vestnik Voronezhskogo gos. un-ta. Ser. Sistemnyi analiz i informatsionnye tekhnologii, 2021, no. 1, 33–53 | MR

[26] M. Kutter, F. Jordan, F. Bossen, “Digital signature of color images using amplitude modulation”, Proc. SPIE, 518–526

[27] J. Zhao, E. Koch, “Embedding robust labels into images for copyright protection”, Proc. Intern. Congress Intellectual Property Rights for Specialized Information, Knowledge and New Technologies (Vienna, August 1995), 242–251

[28] X. P. Zhang, S. Z. Wang, “Efficient steganographic embedding by exploiting modification direction”, IEEE Commun. Let., 10:11 (2006), 781–783 | DOI

[29] G. Paul, I. Davidson, I. Mukherjee, S. S. Ravi, “Keyless dynamic optimal multi-bit image steganography using energetic pixels”, Multimedia Tools Appl., 76 (2017), 7445–7471 | DOI

[30] Digital Data Embedding Laboratory Department of Electrical and Computer Engineering SUNY Binghamton, http://dde.binghamton.edu/download/stego_algorithms/