Passive copy move forgery detection using SURF, HOG and SIFT features

Abstract

Copy-Move in an image might be done to duplicate something or to hide an undesirable region. So in this paper we propose a novel method to detect copy-move forgery detection (CMFD) using Speed-Up Robust Features (SURF), Histogram Oriented Gradient (HOG) and Scale Invariant Features Transform (SIFT), image features. SIFT and SURF image features are immune to various transformations like rotation, scaling, translation etc., so SIFT and SURF image features help in detecting Copy-Move regions more accurately in compared to other image features. We have compared our method for different features and SIFT features show better results among them. For enhancement of performance and complete localization to Copy Move region, a hybrid SURF-HOG and SIFT-HOG features are considered for CMFD. We are getting commendable results for CMFD using hybrid features. © Springer International Publishing Switzerland 2015.