TY - GEN
T1 - Copy-Move Forgery Detection Using Segment Anything Model and BRISK Feature
AU - Akbar, Muhammad Tiyas Fachreza
AU - Studiawan, Hudan
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - In today's age., the image manipulation techniques make it more challenging to verify the authenticity of digital content. Copy-move forgery., where objects of an image are copied and pasted within the same image., is a common type of manipulation that generally used. This research presents an approach for copy-move forgery detection (CMFD). It combines the Segment Anything Model (SAM) with the Binary Robust Invariant Scalable Keypoints (BRISK) feature extraction technique. It applies the SAM's segmentation ability to provide a robust foundation for detecting image components., and BRISK efficiency and reliability for keypoint detection and description. The results implies that the method has a well-balanced detection capability and effectively managing the trade-off between false positives and false negatives. It shown in the performance metrics with accuracy of 92 %., precision of 91 %., recall of 93%., and f1 score of 92%. The integration of SAM and BRISK enhances the robustness of the CMFD and makes it adaptable to difference image types and resilient to common image transformations like rotating and scaling the objects shown by TPR of 87.5% for the rotation of 30 degrees transformation., and 97.4% for the rotation of 30 degrees and scaled for 120%.
AB - In today's age., the image manipulation techniques make it more challenging to verify the authenticity of digital content. Copy-move forgery., where objects of an image are copied and pasted within the same image., is a common type of manipulation that generally used. This research presents an approach for copy-move forgery detection (CMFD). It combines the Segment Anything Model (SAM) with the Binary Robust Invariant Scalable Keypoints (BRISK) feature extraction technique. It applies the SAM's segmentation ability to provide a robust foundation for detecting image components., and BRISK efficiency and reliability for keypoint detection and description. The results implies that the method has a well-balanced detection capability and effectively managing the trade-off between false positives and false negatives. It shown in the performance metrics with accuracy of 92 %., precision of 91 %., recall of 93%., and f1 score of 92%. The integration of SAM and BRISK enhances the robustness of the CMFD and makes it adaptable to difference image types and resilient to common image transformations like rotating and scaling the objects shown by TPR of 87.5% for the rotation of 30 degrees transformation., and 97.4% for the rotation of 30 degrees and scaled for 120%.
KW - BRISK feature
KW - copy-move forgery
KW - image forgery detection
KW - segment anything model
UR - http://www.scopus.com/inward/record.url?scp=85207482378&partnerID=8YFLogxK
U2 - 10.1109/ICSCC62041.2024.10690726
DO - 10.1109/ICSCC62041.2024.10690726
M3 - Conference contribution
AN - SCOPUS:85207482378
T3 - 2024 10th International Conference on Smart Computing and Communication, ICSCC 2024
SP - 555
EP - 559
BT - 2024 10th International Conference on Smart Computing and Communication, ICSCC 2024
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 10th International Conference on Smart Computing and Communication, ICSCC 2024
Y2 - 25 July 2024 through 27 July 2024
ER -