UNCOVER Project Presents a Novel Approach to JPEG Steganalysis at WIFS 2023

UNCOVER Project Presents a Novel Approach to JPEG Steganalysis at WIFS 2023

The UNCOVER project, which aims to develop a framework for uncovering hidden data in digital media, participated in the IEEE International Workshop on Information Forensics and Security (WIFS) in Nuremberg, Germany, on December 4-7, 2023. The workshop is a premier event for researchers and practitioners working on information security and forensics.

The UNCOVER project experts presented a paper titled “JPEG Steganalysis Using Leaked Cover Thumbnails”, which proposes a novel approach to detect steganography in JPEG images. Steganography is the art of hiding information within media files, such as images, audio, video, or text. It poses a challenge for forensic investigations, as it can be used to conceal malware, messages, or other illicit data.

The paper, authored by Martin Benes, Benedikt Lorch and Rainer Bohme from University of Innsbruck, shows how operational steganalysis can be facilitated by acquiring additional information, such as a thumbnail of the cover image found in the seized device, or as an outdated metadata record. Cover thumbnails are small versions of the original images, which are often generated by digital cameras, smartphones, or other devices.

The paper categorizes cover thumbnails into different types: pre-compression and post-compression, depending on whether they are generated from uncompressed pre-cover or compressed cover. The paper quantifies the information left in the cover thumbnails, and demonstrates that thumbnails improve the detector performance, and, in some cases, allow for near-perfect steganalysis by re-creating the thumbnail.

The paper received positive feedback from the audience and sparked a lively discussion. The paper is available online at BLB2023_WIFS.pdf (uibk.ac.at)