Modified SIFT-based Kirsch edge detection approach for copy-move forgery detection

Copy-move forgery (CMF) is one of the most common and challenging image forgeries due to its seamless duplication. This paper proposes a passive detection method that combines a modified Kirsch (mKirsch) edge detector with a novel SIFT-based descriptor (DivSIFT) to effectively identify and localize CMF. The mKirsch detector enhances edge sensitivity by removing specific directional masks, boosting the quality of keypoints extracted by DivSIFT. Experiments were conducted on three benchmark datasets; MICC-F220, CoMoFoD, and MICC-F8Multi under various attack conditions including rotation, scaling, JPEG compression, and multiple cloning. The proposed method achieved high accuracy, notably reaching a 90.91% true positive rate (TPR), 100% precision, and a 95.24% F-measure when NE_SE or SW_NW masks were removed. It also maintained robustness under rotation (81.82% TPR) and scaling (96.97% TPR). Compared to state-of-the-art methods, our approach achieved a lower false positive rate (0%) and faster execution time (2.74 seconds), demonstrating its practical value in real-world forensics.

Text 126296.pdf - Published Version Available under License Creative Commons Attribution Share Alike. Download (8MB) Official URL or Download Paper: https://ascijournal.eu/index.php/asci/article/view...

Actions (login required)

View Item