University of Delaware - College of Engineering
ELECTRICAL & COMPUTER ENGINEERING

Research

Communication & Signal Processing

Steganalysis of Digital Images and Videos

Charles Boncelet

Steganalysis is the search for hidden messages in otherwise innocuous media. For instance, a traitor or criminal might embed secret or proprietary information in a digital image and place that image on a publicly accessible website. Anyone can download that image, but only a confederate with knowledge of a secret key could recover the hidden message. Indeed, most users would be unaware that a hidden message exists.

In our work, we are developing methods that can detect the presence of hidden messages in images and videos. The central idea is to use image models developed in lossless image compressors and use these models to identify anomalies.

Work to date includes excellent results on both least significant bit embedding and +-1 embedding. The main emphasis of our current work is in extending these results to DCT encoded images, such as those in JPEG format.

Recent publications

"A Performance Study of +-1 Steganalysis Employing a Realistic Operating Scenario," L. Marvel, B. Henz, and C. G. Boncelet Jr., Proceedings of 2007 MILCOM, October 2007.

"Steganalysis of +-1 Embedding using Lossless Image Compression," C. G. Boncelet Jr.and L. Marvel, Proceedings of the 2007 IEEE International Conference on Image Processing, September 2007.

"Compression-Based Steganalysis of LSB Embedded Images," C. G. Boncelet Jr., L. M. Marvel, and A. Raglin, Proceedings of Electronic Imaging'06, Jan 2006.


Bookmark and Share