Copy-Move Image Forgery Detection Using DCT and ORB Feature Set

Abstract

The unprecedented use of digital images and videos for communication, the ease of access and use of graphic editing applications have consequently led to the increased importance of detecting copy-move forgery. The proposed copy-move forgery detection (CMFD) technique relies on DCT and ORB feature extraction and distance-based clustering approach. Extracted DCT features are matched based on Euclidean distance. Extracted key-points using ORB are matched using k-NN procedure based on Hamming distances. To improve accuracy, false matches are removed with the help of a distance-based clustering technique. The proposed technique is applied for testing on CoMoFoD small dataset. Results on experimentation showcase that the technique is efficient in detecting copy-move forged regions and also robust towards brightness and contrast change, noise addition, geometric transformations like scaling and rotation and several forgeries. The proposed technique is compared with two state-of-the art techniques. © 2020, Springer Nature Singapore Pte Ltd.