Copy-Move Image Forgery Detection and Classification Using Sift and Birch Approaches

College of Big Data and Internet, Shenzhen Technology University, 3002 Lantian Road, Pingshan District, Shenzhen, Guangdong, 518118, China

This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Copy-move forgery detection (CMFD) continues to pose significant challenges in digital image forensics, particularly under geometric transformations such as rotation and scaling. This study presents a detection pipeline that couples Scale-Invariant Feature Transform (SIFT) keypoints with the Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm. Leveraging BIRCH's robustness to noise, computational efficiency and relatively mild hyper-parameter sensitivity, the proposed framework effectively suppresses spurious correspondences and isolates candidate tampered regions with high confidence. Performance is evaluated on the MICC-F220 benchmark and benchmarked against two alternative clustering strategies-Hierarchical Agglomerative Clustering (HAC) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN)-each configured with their best-reported parameter sets. Under the tested settings, the SIFT + BIRCH scheme attains a true-positive rate of 99 % while sustaining a low false-positive rate, outperforming both baselines. The results underscore the suitability of hierarchical clustering for spatially grouping SIFT keypoints and highlight BIRCH as a promising direction for future CMFD research, particularly when integrated with complementary outlier-rejection mechanisms.