Blind forensics of image gamma transformation and its application in splicing detection

Abstract

Image forensics technology based on image manipulation detection has aroused great interest of researchers in recent decades. By revealing the traces of image manipulations, it could contribute to the location of tampered areas. This paper addresses the detection of image gamma transformation and its application in image splicing detection. A 5-dimensional feature vector is constructed based on the effects of gamma transformation on the histogram, and then Support Vector Machine (SVM) is trained to detect gamma transformation. For splicing tampered area locating, the investigated image is divided into overlapping blocks based on the sliding window, and then each pixel gets a probability of being gamma transformed according to the detection results in image blocks, based on which the tampered area is located. The experimental results validate the effectiveness of the proposed method when no post-operations are applied. Meanwhile, the proposed method is robust to the attack of rank filtering.