Steganalysis attacks on stego-images using stego-signatures and statistical image properties

This paper discusses a strategy for detecting hidden messages in stego-images in Internet traffic. The strategy adopts attacks that are based on both the signatures of the most popular steganographic techniques and some of the statistical parameters of the images being analyzed. A gradual approach for the attacks is applied since the speed with which detection is to be achieved is an important factor. An extensive set of images with different file format including JPEG, GIF and BMP, different cover and stego object sizes was processed and the results achieved indicate a high degree of success.