StegKit: Automated Steganalysis Tool

Surreptitious communication through steganography has application in many settings. There are many different techniques for hiding steganographic content and in theory, any file type can serve as a carrier. Most studies in steganalysis have focused on single file types and a few steganographic methods on those file types, representing a limited sample of the population of transmitted data. We propose to develop of a general-purpose steganalysis system that will handle many file types, many steganographic methods, while being easy to configure and use. Such a system can be deployed widely enough to adequately sample a statistically significant portion of transmitted data. A distinguishing feature of our approach is the use of machine learning techniques to create a file type classifier that can be used to select among available steganalysis modules. The resulting system will be modeled on modern antivirus software, capable of background monitoring of files, email, and visited web pages.