**3.1. Algorithm based on invariant keypoints**

In contrast to other algorithms, this algorithm does not divide the image into auction blocks to extract features but instead extracts features from the intact image. Feature extraction is performed with SIFT and speeded-up robust feature (SURF). This technique is applied to derive the characteristic local feature of an image and produce a keypoint in accordance with preset requirements. The vector sum/descry values are fixed for rotational, translational, and scale measurements and are partially fixed for strong illumination changes in local geometric distortion [14, 15]. The first attempt to exploit this algorithm was reported by [16]. In the algorithms, only the correspondence of the keypoint can be achieved by its maximum bin, including the identity of the nearest neighbor [17]. SIFT has been adopted to identify replicated regions in a counterfeit image. The SIFT signifier is applied to detect copied areas by coping with keypoints rather than clusters. This algorithm has excellent detection accuracy but otherwise poor performance.
