Statistic test of hidden data existence in digital image and audio signals

As an important aspect in the study of attacks against information hiding and the development of anti-attack techniques, this paper describes a method to compare a pair of perceptibly identical digital images or audio signals in order to single out the one that contains embedded hidden-information. By means of statistic analyses and comparison between the two signals, and making use of short-term correlativity in the space/time domain, three criteria are proposed, which are mutually complementary, and can be used in statistical tests to identify signals with embedded data. The experimental results are satisfactory.