ICASSP 2020accepted0 citations
Selection-Channel-Aware Reverse JPEG Compatibility for Highly Reliable Steganalysis of JPEG Images
Abstract
This paper deeply studies the principle of the recent reverse JPEG compatibility attack [1]. This analysis allows us to cast the problem of hidden data detection in DCT coefficients within hypothesis testing theory. The optimal LR test, thought efficient, is rather computationally expensive. Therefore, mild assumptions are used to simplify the detection problem dramatically and design a test that is simple yet extremely efficient and reliable. It is shown that the proposed detector is way more efficient than the original statistical test [1], and allows highly reliable detection of data hidden within JPEG images.
BibTeX
@inproceedings{icassp2020_selectionchannel,
title = {Selection-Channel-Aware Reverse JPEG Compatibility for Highly Reliable Steganalysis of JPEG Images},
author = {Rémi Cogranne},
booktitle = {ICASSP 2020},
year = {2020}
}