Research of main components of machine learning based JPEG-stegananalysis systems

Authors

  • Наталія Василівна Кошкіна V.M. Glushkov NAS of Ukraine

DOI:

https://doi.org/10.18372/2410-7840.22.14801

Keywords:

information security, steganalysis, passive counteraction (steganalysis), machine learning methods, comparative analysis, features based models, SVM, ensemble classifier

Abstract

To create effective steganalysis systems in the given practical conditions it is necessary to perform analysis and quality estimation of existing methods and components. To select optimal system components it is required to compare the estimates of basic characteristics of the candidates available. However, making such comparison based on a data from scientific publications is quite difficult due to differences of the conditions of numerical experiments. The basis of this study is the principle of creating equal conditions for all the investigated statistical features based models for JPEG steganalysis using machine-learning methods. We analyzed the detection performance and accuracy for four different variants of data hiding in the frequency domain that were obtained using statistical models such as CHEN, CC-CHEN, LIU, CC-PEV, CC-C300, GFR and DCTR, as well as SVM with linear or Gaussian kernel functions or ensemble classifier. The main results of the study are tables containing numerical estimates of performance of the main stages of steganoanalysis and the classification accuracy of empty and stego images. Model LIU was the slowest for extraction of features vector, but it is the fastest in the classifier training process. Models CC-PEV, LIU and DCTR combined with the ensemble classifier provided the best detection accuracy for stego images created by programs like Jsteg, Jphide and Steganos Privacy Suite. The detection accuracy of stego images by Jsteg, Jphide and Steganos in our experiments reached 100, 95.3 and 99.8% accordingly. J-UNIWARD method has proved itself to be more secure than LSB-steganography. The GFR and DCTR models are proved to be the most sensitive to J-UNIWARD transformation. In distinction to other investigated models they analyze statistics in the spatial domain of the image but not in the frequency domain - where the message was embedded.

Author Biography

Наталія Василівна Кошкіна, V.M. Glushkov NAS of Ukraine

Doctor of Engineering Sciences (Information security), Senior Researcher, Department of Numerical Methods Optimization, V.M. Glushkov NAS of Ukraine.

References

Н.В. Кошкина, “Обзор и классификация методов стеганоанализа”, Управляющие системы и машины, № 3, с. 3-12, 2015.

T. Filler, J. Judas, J. Fridrich “Minimizing Embedding Impact in Steganography using Trellis-Coded Quantization”, Proc. SPIE, Electronic Imaging, Media Forensics and Security XII, San Jose, CA, vol. 7541, 2010.

C. Chen, Y.Q. Shi, “JPEG image steganalysis utilizing both intrablock and interblock correlations”, IEEE ISCAS, International Symposium on Circuits and Systems, рр. 3029-3032, 2008.

J. Kodovsky, J. Fridrich, “Calibration revisited”, In J. Dittmann,S. Craver, and J. Fridrich, editors, Proceedings of the 11th ACM Multimedia and Security Workshop, рр. 63-74, 2009.

Q. Liu, “Steganalysis of DCT–embedding based adaptive steganography and YASS”, In J. Dittmann, S. Craver, and C. Heitzenrater, editors, Proceedings of the 13th ACM Multimedia & Security Workshop, рр. 77-86, 2011.

T. Pevny, J. Fridrich, “Merging Markov and DCT features for multiclass JPEG steganalysis”, In E. J. Delp, P. W. Wong, editors, Proceedings SPIE, Electronic Imaging, Security, Steganography, and Watermarking of Multimedia Contents IX, vol. 6505, pp. 301-314, 2007.

J. Kodovsky, J. Fridrich, “Steganalysis in high dimensions: fusing classifiers built on random subspaces”, 8th SPIE Electronic Imaging, Media, Watermarking, Security and Forensics, vol. 7880, pp. 1-13, 2011.

X. Song, F. Liu, C. Yang, X. Luo, Y. Zhang, “Steganalysis of Adaptive JPEG Steganography Using 2D Gabor Filters”, Proceedings of the 3rd ACM Workshop on Information Hiding and Multimedia Security. ACM, pp.15-23, 2015.

V. Holub, J. Fridrich, “Low Complexity Features for JPEG Steganalysis Using Undecimated DCT”, IEEE Transactions on Information Forensics and Security, vol. 10, no. 2, pp. 219-228.

J.Kodovský, J. Fridrich, V. Holub, “Ensemble Classifiers for Steganalysis of Digital Media”, IEEE Transactions on Information Forensics and Security, vol. 7, no. 2, pp. 432-444, 2012.

Downloads

Published

2020-07-01

Issue

Vol. 22 No. 2 (2020)

Section

Articles

License

Authors who publish with this journal agree to the following terms:

  1. Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.

  2. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.

  3. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).