Image authentication by statistical analysis

Abstract

This paper investigates the discrimination between Photographic Images (PIM) and Computer Generated (CG) images. The proposed method exploits traces of Color Filter Array (CFA) interpolation, present in PIM images, together with the use of hypothesis testing theory. By using the Likelihood Ratio Test (LRT), the method proposed to distinguish PIM from CG images warrants a prescribed False Alarm Rate (FAR) and achieves the maximal detection power. Experimental results show the efficiency of the proposed methodology and the high robustness with respect to anti-forensic techniques.