A new approach for localization of copy-move forgery in digital images

In this study, newly localization approach is presented to reveal copy-move forgery regions. Firstly, Speed-Up Robust Features (SURF) keypoints are utilized to label whether the suspicious input image is original or forged. With the presence of a sufficient number of the most similar keypoint matches, the image is labeled as forged and the location stage of the forged region is started. In determining the boundaries of the forged region, the blocks received around the matched keypoints are considered seed forged blocks. 8 overlapping neighboring blocks of these blocks in the target and source regions are labeled as candidate blocks. If the candidate blocks corresponding to each other meet the similarity requirement, they are labeled as forged blocks and the candidate block label is removed. The location phase is completed by evaluating the candidate blocks located in the neighborhood of all forged blocks. The high performance of the method has been proved in the freely available GRIP dataset with the experiments.