Passive Copy-Move Forgery Detection using SIFT, HOG and SURF Features

Abstract

Copy-move is a common type of digital image forgery. In an image, Copy-Move tampering might be done to hide an undesirable region, or to duplicate something in the image. These images might be used for necessary purpose like evidence in the court of law. So, authenticity verification plays a vital role for digital images. In this paper, we compare the CMFD (Copy-Move Forgery Detection) using Image features like SIFT (Scale Invariant Features Transform), HOG (Histogram Oriented Gradient) and SURF (Speed-Up Robust Features) and hybrid features (SURF-HOG and SIFT-HOG). The comparison results show that CMFD using SIFT features provide better results as compared with SURF and HOG features. Also, considering hybrid features, SIFT-HOG and SURF-HOG produce better results for CMFD using SIFT, SURF or HOG alone.

Keywords: Passive Forensics, Copy-Move Forgery Detection, Single Region CMFD, Multiple Region CMFD, Histogram Oriented Gradient, Scale Invariant Features Transform, Speed-Up Robust Features.