A jump patch-block match algorithm for multiple forgery detection

Abstract

The Image analysis part of Digital Forensics plays an important role in investigating tampering of images. Exemplar based inpainting and copy paste forgeries play visual tricks to deceive people and affect the authenticity of the images. We have devised a blind detection method which is based on the luminance component of the image and median comparison of the blocks in the Region of Suspicion (ROS). The median comparison of the blocks facilitates the “Jump patch” functionality of the proposed method and hence makes this method robust and faster than the already existing methods. Detecting two forgeries by a single method makes it unique. A number of tampered natural images are used to show the efficacy of the algorithm.