Priming for Steganography: Undermining Steganalysis through the Introduction of Superfluous Variation

This paper examines the effects upon steganalysis classifier performance of introducing superfluous variation into digital cover images . By using fixed steganography and steganalysis tools (Steghide 0.5.1 for former and Stegdetect 0.6 for latter), a number of different image filters are individually applied to cover images prior to the introduction of steganographic content; the effect of these filters are then measured in terms of the classifier performance (AUC) of steganalysis software Stegdetect.