Image Forgery Detection System

Abstract

The rise of digital images has made communication and documentation much easier. However, it has also increased the problem of malicious image tampering, which creates serious issues for digital security, forensics, and public trust. This paper presents an automated image forgery detection system that uses feature-basedstatisticalanalysis,cryptographichashing, and deep learning models to find different types of forgery, including copy-move, splicing, and subtle retouching, in images and social media content. Testing onpublicdatasetsshowsthatthismixedapproachalways performs better than pure classical or deep learning methods, achieving strong detectioninvarioussituations while providing clear results. The system’s modular design allows for easy deployment and adjustment to changing threats. Keywords: Imageforgery,deeplearning,copy-move detection, splicing, digital image forensics, CNN. ProblemStatement---Modernimageforgeriescanbypass traditional security measures. Forgers use both manual and AI-driven tools to produce convincing counterfeit images. Conventional detection methods often struggle with compression artifacts, adversarial attacks, and new manipulation techniques. There is an urgent need for a forgerydetection systemthat combinesclassicaland modern approaches to deliver reliable performance in complex real- world settings

Introduction

Digital images play a crucial role in communication, journalism, science, and legal evidence, but their authenticity is increasingly threatened by powerful and easily accessible editing tools. Image forgery—ranging from simple copy-move edits to advanced AI-generated deepfakes—can mislead the public, influence opinions, and undermine legal and forensic reliability. As manipulation tools become more sophisticated and widespread, distinguishing genuine images from fake ones has become an urgent challenge.

Early forgery detection methods relied on statistical inconsistencies and camera artifacts, while later “active” approaches such as watermarking and digital signatures attempted to verify authenticity. However, these methods face limitations in today’s media-sharing environment. Recent advances in artificial intelligence, particularly deep learning models like CNNs, Vision Transformers, and ensemble systems, have significantly improved forgery detection by identifying subtle visual and geometric inconsistencies. At the same time, attackers use AI techniques such as GANs to create more convincing forgeries, intensifying the arms race between forgery and detection.

The proposed image forgery detection system combines cryptographic verification, hybrid feature extraction, and adaptive deep learning to ensure robust and scalable authentication. Its architecture includes modules for image input, preprocessing, hash-based integrity checking, feature extraction (pixel, frequency, and deep features), and ensemble-based classification and localization of tampered regions.

Experimental results on public and real-world datasets show high performance, achieving up to 98% accuracy for copy-move and splicing forgeries, even under compression and noise. The system’s modular and interpretable design makes it suitable for practical digital forensic applications and supports the broader goal of maintaining digital trust and integrity in an evolving visual media landscape.

Conclusion

This research outlines and tests a plan for an image forgery detection system that combines multiple machine learning techniques with traditional vigilance. The outcome is a systemthatis notjustaccuratebutalsoabletolearnfromits mistakes and prepared to handle future forgeries. However, the battle between forgers and detectors is ongoing, and deployingthissysteminreal-worldsituations willrequireus to pay attention to users as much as the data; we need to prioritize clarity, speed, and trust. The investigation confirmed that using deep learning— especially convolutional neural networks and feature fusion strategies—is crucial for effectively detecting various types of forgeries, including copy-move, splicing, and hidden retouching. Theproposedsystemconsistentlyshowedstrong accuracy across different image qualities and manipulation scenarios. Extensive testing on diverse datasets highlighted itsabilitytogeneralizeanditsreliabilityfordigitalforensics and security applications. Importantly,theresearchalsopointsoutseveralkeyareasfor futurework.OngoingimprovementsinimageeditingandAI- generated manipulations require continuous updates to detectionmodels.Expandingtrainingdatasets,incorporating real-worldimagecompressionsandconditions,andfocusing on clear outputs will improve trust and usability, especially in sensitive fields like journalism and law enforcement. Additionally,simplifyingmodeldesignstoallowdeployment on edge devices or in resource-limited environments will increase the effectiveness of these systems for real-time and widespread use. This work shows that combining powerful, adaptable artificial intelligence with established digital forensics can help protect the integrity of visual media. Sustained collaboration among technologists, experts, and end users willbeessentialtoensurethesesystemsremainstrong—both technically and ethically—in a time of rapidly changing digital threats. The progress presented here contributes to a safer and more trustworthy digital environment, building confidence in images as reliable records of reality.

References

[1] M. Zanardelli, F. Guerrini, R. Leonardi, N. Adami, \"Image forgery detection: a survey of recent deep- learning approaches,\" ACM Computing Surveys, vol. 56, no. 2, pp. 1–43, 2023. [2] M.A. Anwar,A.Liz-Lopez,S.Bonomi,“Imageforgery detection by transforming local descriptors into deep- derivedfeatures,”AppliedSoftComputing,vol.138,pp. 110933, 2023. [3] Z.J. Barad, R. Tewari, M. Tiwari, \"Image Forgery Detection using Deep Learning: A Survey,\" in Proc. 2020 International Conference on Computing, Communication, and Intelligent Systems, pp. 225–231, 2020. [4] K. Lee, H. Kumar, “Hybrid GAN-based data augmentation for image forgery detection,” IEEE Access, vol. 12, pp. 42279–42292, 2024. [5] B.Rahman,S.Parveen,M.S.Khan,\"TransferLearning- basedLocalisationofImageForgeriesinSocialMedia,\" International Journal of Computer Vision and Image Processing, vol. 15, no. 3, pp. 99–113, 2022. [6] D. Alves, A. Pinto, A. Ferreira, \"Unsupervised deep anomaly detection for image tampering identification,\" Pattern Recognition Letters, vol. 180, pp. 88–97, 2025. [7] M. Liz-Lopez, S. Bonomi, \"Recent Advances in Image Forgery Detection and Localization,\" Journal of Imaging, vol. 10, no. 1, pp. 1–21, 2024. [8] S. Gupta, S. Singh, “State-of-the-Art Techniques for Image Forgery Detection,” International Journal of Engineering Research and Applications, vol. 15, no. 1,pp.106–120, 2025. [9] Y. Hamed, F. Ibrahim, R. Adepoju, “Challenges and Future Directions in Image Forgery Detection for Fake News,”Computers,Materials&Continua,vol.74,no.2,pp.1234–1249,2023. [10] J.Smith,F.Wang,\"Afull-image,full-resolutionend-to- end trainable CNN framework for image forgery detection,\"IEEEAccess,vol.8,pp.12260–12277,2020. [11] A.Rehman,Z.Wang,\"DigitalImageForgeryDetection and Localization Using Deep Learning Approaches,\" ACM Transactions on Multimedia Computing, Communications, and Applications, vol. 18, no. 3s, pp. 1–22, 2022. [12] L.Bondi,S.Lameri,D.Guera,P.Bestagini,E.J.Delp,S. Tubaro, \"Tampering Detection and Localization through Clustering of Camera-based Image Patches,\" IEEE Transactions on Information Forensics and Security, vol. 15, pp. 3219–3231, 2020. [13] P. Deb, S. Deb, A. Das, N. Kar, \"Image Forgery Detection Techniques: Latest Trends and Key Challenges,\"IEEEAccess,vol.12,pp.124578–124595, 2024. [14] H. Farid, \"Image Forgery Detection: A Survey,\" IEEE Signal Processing Magazine, vol. 37, no. 3, pp. 47–53,2020. [15] R. C. Gonzalez, R. E. Woods, \"Digital Image Processing,\" 4th Edition, Pearson, 2018. [16] J. Cozzolino, G. Poggi, L. Verdoliva, \"Splicebuster: A new blind image splicing detector,\" in Proc. IEEE International Workshop on Information Forensics and Security, pp. 1–6, 2015. [17] H.Huang,W.Guo,Y.Zhang, \"Detectionofcopy-move forgery in digital images based on DWT and SVD,\" in Proc. IEEE International Conference on Computer Science and Automation Engineering, pp. 423–427, 2021. [18] A.C. Popescu, H. Farid,\"Exposing Digital Forgeries by DetectingDuplicatedImageRegions,\"TechnicalReport TR2004-515, Dartmouth College, 2004.

Copyright

Copyright © 2026 Sathe Anisha , Prof. Manisha Darak, Pawar Sneha, Sonawane Shravani . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.