Matlab implementation of the framework for quantitative steganalysis in high-dimensional feature spaces as proposed in [1]. The algorithm is based on gradient boosting [2] and utilizes different random subspace at each stage.
The main function, GBM_steganalysis.m is accompanied by an example script example.m.
In this paper, we propose a regression framework for steganalysis of digital images that utilizes the recently proposed rich models -- high-dimensional statistical image descriptors that have been shown to substantially improve classical (binary) steganalysis. Our proposed system is based on gradient boosting and utilizes a steganalysis-specific variant of regression trees as base learners. The conducted experiments confirm that the proposed system outperforms prior quantitative steganalysis (both structural and feature-based) across a wide range of steganographic schemes: HUGO, LSB replacement, nsF5, BCHopt, and MME3.
[1] J. Kodovský and J. Fridrich, Quantitative Steganalysis Using Rich Models, SPIE, Electronic Imaging, Media Watermarking, Security, and Forensics XV, San Francisco, CA, February 3-7, 2013.
[2] J. H. Friedman, Greedy function approximation: A gradient boosting machine, Annals of Statistics, 29:1189–1232, 2000.