ACO-Enhanced Siamese Networks for Robust Feature Matching in Copy-Move Image Forgery Detection

Copy-move forgery detection (CMFD) is a critical task in image forensics, aiming to identify tampered regions that are duplicated and manipulated within an image. Traditional methods often struggle with handling complex transformations such as scaling, rotation, and noise addition, limiting their effectiveness in real-world applications. This study proposes a novel approach utilizing Ant Colony Optimization (ACO)-Enhanced Siamese Networks for robust feature matching in copy-move forgery detection. The methodology integrates the strength of Siamese Networks in extracting and comparing patch-based image features with the optimization capabilities of ACO to refine the feature-matching process. The Siamese Network is designed to process overlapping patches, generating embeddings that are used for similarity assessment. ACO is employed to optimize the selection of key feature points and minimize redundant computations, ensuring accurate and efficient forgery detection. The proposed framework also incorporates pre-processing techniques to normalize input images and address geometric transformations commonly seen in forgery cases. Extensive experiments were conducted on benchmark datasets, evaluating the model against state-of-the-art techniques. The proposed model achieved an accuracy of **97.13%**, significantly outperforming existing methods, particularly in scenarios involving challenging image transformations. The results highlight the effectiveness of combining Siamese Networks with ACO for enhanced feature matching in CMFD, offering a scalable and efficient solution for digital image forensics. This approach not only improves detection accuracy but also reduces computational overhead, making it suitable for real-time forensic applications. The proposed methodology represents a significant advancement in the field, addressing key limitations of existing techniques while paving the way for future research in automated forgery detection.