Correlation analysis for illegal tampering image evidence detection

Abstract

Recently, forged pictures, pirated image products, and tampered photos become more and more rampant in our society, which may endanger the rights of owners to earn illegal profits or change the key evidence in law case. It is urgent to accurately trace the real source of images by using the digital image forensics technology. This paper adopts correlation coefficient according to the image content for digital image tampering detection. The image to be detected is masked by a sliding window which moves from left to right or up to bottom. The area in the window is matched with the fingerprint, which is related to the used camera, to determine the tampered area in the image. The fingerprint is determined by the photo‐response non‐uniformity (PRNU) sensor. It is estimated by the residue noise and a group of images from the camera. The correlation coefficient based method can detect the tampered areas from the same image or several different images. The experiments on synthetic tampered images demonstrate most of the tampered areas can be detected.