We consider methods for answering reliably the question of whether an image contains hidden data; the focus is on grayscale bitmap images and simple LSB steganography. Using a distributed computation network and a library of over 30,000 images we have been carefully evaluating the reliability of various steganalysis methods. The results suggest a number of improvements to the standard techiques, with particular benefits gained by not attempting to estimate the hidden message length. Extensive experimentation shows that the improved methods allow reliable detection of LSB steganography with between 2 and 6 times smaller embedded messages.
Andrew D. Ker