Steganography and Steganalysis of Raw and Compressed Image Data
thesisposted on 2019-12-01, 00:00 authored by Mehdi Sharifzadeh
Steganography is the art of hiding data in a cover medium without arousing suspicion of the warden. In this thesis, we focus on the most popular and studied cover medium for steganography, digital images. In digital image steganography, the statistical model of an image is essential for hiding data in less detectable regions and achieving better security. This has been addressed in the literature, where different cost-based and statistical model-based approaches were proposed. However, due to the usage of heuristically defined distortions or statistical models resulting in numerically solvable equations, there is no closed-form expression for security as a function of payload. The closed-form expression is crucial for a better insight into image steganography, batch steganography, and pool steganalysis problems. Besides, it is also required for improving the security of steganography and batch steganography algorithms against single image and pool steganalysis. Towards this goal, our research is focused on four problems. 1) We develop a general spatial image steganography embedding model that can utilize embedding costs and residual variances for embedding the hidden message and achieves state-of-the-art performance. 2) We extend the embedding model to JPEG steganography, which is also generalized in the sense that it can accomplish embedding using any spatial or DCT embedding cost as well as residual variances. Employing the proposed model improves the security of previous works and outperforms the state-of-the-art JPEG steganography algorithms. 3) We derive the closed-form expression for steganalysis error of batch steganography. The expression allows us to study the effect of batch size on security which results in a novel batch steganography method, Adaptive Batch size Image Merging steganographer (AdaBIM). 4) We further extend the closed-form expression of single image steganalysis to pool steganalysis for an optimal omniscience detector. The developed analytical model is validated by its ability to accurately estimate empirical results of pool steganalysis and predict the behavior of empirical pool steganalysis error variance.
DepartmentElectrical and Computer engineering
Degree GrantorUniversity of Illinois at Chicago
Degree namePhD, Doctor of Philosophy
Committee MemberAnsari, Rashid Seferoglu, Hulya Soltanalian, Mojtaba Ziebart, Brian
Submitted dateDecember 2019