Passive Detection of Tampered JPEG Image Based on First Digit Statistics

Abstract

Most of the time, some traces of digital tampering may be left in image manipulation, resulting in the inconsistency of natural images and providing some clues for image forgeries. This paper proposed a novel probability model based on the first digit statistics of DCT coefficients, to express the changes of statistical properties after manipulation. And we combined Bayes' theorem to detect and locate the tampered regions. Experimental results show that our method can exactly locate the tampered regions, and the performance is better than the existing algorithms especially when the first compression quality factor is larger than the second one. It is also proved that this method is resistant to different kinds of forgery techniques.