Automated Steganography Detection For Law Enforcement Continuation

Steganography (steg) is a growing concern for law enforcement because it compounds issues posed by encrypted evidence with the complicating factor that evidence is hidden behind an innocuous façade (the carrier). The University of Rhode Island (URI) has developed blind steg detection for JPEG and MP3 carriers using sophisticated machine learning/classification techniques, including Support Vector Machines and Linear Discriminant Analysis, which have yielded very accurate results. The detection software has been integrated in the EnCase computer forensic analysis tool, and with a web-based interface for uploading candidate files for detection. This project proposes three important extensions to URI's previous work:
' Integration of the steg detection algorithms with the Wetstone Technologies Stego Suite ' the leading steg detection and breaking tool suite used by law enforcement;
' Generalization of detection techniques to enable creating steg detection for new steg
programs as those programs come out;
' Development of steg detection for video ' an emerging problem area for which there is
no current solution.

The deliverables will be new steg detection algorithms functioning in Stego Suite, a process for easily creating similar algorithms for new steg tools, and techniques for steg detection in video.

ca/ncf