An improved copy-move forgery detection algorithm

Abstract

Advance in technology and editing tools yields to easy manipulation in images. Nowadays, digital images usage is very common thus the importance of detection of forgeries to assure the authenticity of images. Since image tampering are hard to be noticed by the human eye, authenticity of an unknown source image is nearly impossible to detect. Copymove forgery is a very common type of digital image forgery where a part of an image is copied and pasted in another part in the same image for the purpose of hiding or adding something to the image. This paper deals with this type of forgeries and presents several detection methods such as SIFT and SURF. The offered detection is robust to rotating, scaling, horizontal and vertical translation and noise. Moreover, our algorithm can deal with multiple cloning cases. On the whole, the methods presented in this paper proved to be efficient with a rate of 99.53% (for SIFT) and 97.77% (for SURF) for a random set used in [11] of 220 images. Index terms: forgery, SIFT, detector, copy-move, authenticity, feature, matching, clusters.