Efficient steganography detection by means of~compression-based integral classifier

V. A. Monarev ; A. I. Pestunov

Prikladnaâ diskretnaâ matematika, no. 2 (2018), pp. 59-71.

Résumé

We propose a model of an integral classifier in order to solve the problem of binary steganalysis by means of machine-learning tools more efficiently. The problem of binary steganalysis consists in recognizing whether a given container is empty or contains a certain payload embedded via a certain steganographic algorithm. In steganalysis, such problem is often solved using such machine-learning techniques as the support vector machine and the ensemble classifier. Instead of using a single classifier (as it is done now) which is intended to make an ultimate decision about whether the container is empty or not, the proposed in this paper integral classifier consists of several classifiers and works in such a way that each of them processes only those containers which satisfy a certain condition. Within the proposed model, we develop a compression-based integral classifier which works as follows. The training set of classifiers is splitted into several subsets according to the containers compression rate; then a corresponding number of classifiers are trained, but each classifier is injected only with an ascribed subset. The testing containers are distributed between the classifiers (also according to their compression rate) and the decision about the certain container is made by the chosen classifier. In order to demonstrate the power of the integral classifier, we performed some experiments using the famous de-facto standard images database BOSSbase 1.01 as a source of the containers along with contemporary content-adaptive embedding algorithms HUGO, WOW and S-UNIWARD. Comparison with state-of-the-art results (obtained for the single support vector machine and the ensemble classifier) demonstrated that, depending on the case, the integral classifier allows to decrease the detection error by 0.05–0.16.

Détail

Keywords: steganalysis, detection error, WOW, support vector machine, integral classifier, projected spatial rich model, spatial rich model
Mots-clés : HUGO, UNIWARD, ensemble classifier, compression.

Bibliographie

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

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

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

[4] Borisenko B. B., “Modified Hotelling's chart excluding trend influence and its application for digital watermarks detection”, Prikladnaya Diskretnaya Matematika, 2010, no. 2, 42–58 (in Russian)

[5] Menori M., Munir R., “Blind steganalysis for digital images using support vector machine method”, Proc. IEEE Intern. Symp. Electronics and Smart Devices (ISESD), IEEE, Bandung, Indonesia, 2016, 132–136

[6] Pevný T., Fridrich J., Ker A., “From blind to quantitative steganalysis”, IEEE Trans. Inform. Forensics and Security, 7:2 (2010), 445–454 | DOI

[7] Cogranne R., Denemark T., Fridrich J., “Theoretical model of the FLD ensemble classifier based on hypothesis testing theory”, Proc. 6th IEEE Intern. Workshop on Inform. Forensics and Security (WIFS), IEEE, Atlanta, GA, USA, 2014, 167–172

[8] Schöttle P., Korff S., Böhme R., “Weighted stego-image steganalysis for naive content-adaptive embedding”, Proc. 4th IEEE Intern. Workshop on Inform. Forensics and Security (WIFS), IEEE, Tenerife, Spain, 2012, 193–198

[9] Monarev V. A., Pestunov A. I., “Enhancing steganalysis accuracy via tentative filtering of stego-containers”, Prikladnaya Diskretnaya Matematika, 2016, no. 2, 87–99 (in Russian) | DOI | MR

[10] Boncelet C., Marvel L., Raqlin A., “Lossless compression-based steganalysis of LSB embedded images”, Proc. 41st Ann. Conf. on Inform. Sciences and Systems (CISS), IEEE, Baltimore, MD, USA, 2007, 923–929

[11] Monarev V., Pestunov A., “A new compression-based method for estimating LSB replacement rate in color and grayscale images”, Proc. 7th IEEE Intern. Conf. on Intelligent Inform. Hiding and Multimedia Signal Processing (IIH-MSP), IEEE, Dalian, China, 2011, 57–60

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

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

[14] Holub V., Fridrich J., “Digital image steganography using universal distortion”, Proc. 1st ACM Workshop on Inform. Hiding and Multimedia Security (IHMMSec), ACM, Montpellier, France, 2013, 59–68

[15] Holub V., Fridrich J., “Designing steganographic distortion using directional filters”, Proc. 4th IEEE Intern. Workshop on Inform. Forensics and Security (WIFS), IEEE, Tenerife, Spain, 2012, 234–239

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

[17] Large Text Compression Benchmark, , 2017 http://mattmahoney.net/dc/text.html

[18] scikit-learn: Machine Learning in Python, , 2017 http://scikit-learn.org

[19] Monarev V., Pestunov A., “A known-key scenario for steganalysis and a highly accurate detector within it”, Proc. 10th IEEE Intern. Conf. on Intelligent Inform. Hiding and Multimedia Signal Processing (IIH-MSP), IEEE, Kitakyushu, Japan, 2014, 175–178