Steganalysis Techniques for Documents and Images
Final rept. Jan 2003-Dec 2004
POLYTECHNIC UNIV BROOKLYN NY
Pagination or Media Count:
The objective of the proposed research is to develop novel techniques and algorithms for document and image steganalysis. We developed a new framework for LSB steganalysis that gives better performance then the best of the known techniques. We also developed a new methodology of steganalysis by modeling steganography as an additive noise process and measuring the mean and variance of the stego-signal. We derived a formula that governs all additive steganography. We then illustrated the efficacy of our model using variations on LSB steganography. For binary images, we have made significant progress in developing steganalysis methods to detect marked images when the original cover images are either purely electronic images or printed-scanned images. Our work represents the first effort for detecting steganography in binary images.