ORB for Detecting Copy-Move Regions with Scale and Rotation in Image Forensics

The paper proposes a method to detect the forged regions in image using the Oriented FAST and Rotated BRIEF (ORB). In many previous researches in the field of copy-move forgery detection, algorithms mainly focus on objects or parts which are copied, moved and pasted in another places in the same image with the same size of the original parts or included the rotation sometimes, but the copied regions detection with different scale has not much interested in. By adding an oriented component to FAST and the rotation feature to BRIEF, ORB makes the proposed method more powerful and efficient to detect copy-move regions with both scale and rotation. In addition, the removing non-copied objects by calculating their sharpness improves the accuracy of the detection. The experiment is done on the datasets for copy-move images and some real images with the improved time and high accuracy.