Digital images are increasingly being used as steganographic covers for secret communication. The Least Significant Bit (LSB) encoding is one of the most widely used methods for embedding a message in a digital image. However, the direct application of LSB encoding is vulnerable to steganalysis. For example, RS steganalysis is very efficient in detecting the presence of a message in a digital image and to estimate its approximate size. This paper presents a method robust to RS steganalysis, that makes the presence of a message unnoticeable. The method is based on the application of reversible histogram transformation functions to the image, before and after embedding the secret message. The method was tested on 4 greyscale images, with messages of 10%, 30% and 90% of the maximum embedding size. The proposed method proved to be effective in eluding RS steganalysis for all cases tested.
André R. S. Marçal, Patricia R. Pere