Copy-Move Forgery Detection using DCT and SIFT

Abstract

Digital images are the most prevalent way to spread a message. So the authenticity of images is very essential. But due to advancement of the technology the editing of images has become very effortless. Copy-move forgery is most basic technique to alter an image. In this one part of image is copied, called as snippet, and pasted within same image and most likely post-processing it. Considerable number of algorithms is proposed to detect different post-processing on snippet of image. In this paper novel approach is proposed to detect combination of different post-processing operations by single method. It is analyzed that block-based features method DCT is robust to Gaussian noise and JPEG compression, secondly the keypoint-based feature method SIFT is robust to rotation and scaling. Thus by combining SIFT and DCT we are able to detect forgery under post-processing operations of rotation, scaling, Gaussian noise, and JPEG compression and thus the efficiency to detect forgery improves.