Hybrid Deep Feature Fusion Framework for Intelligent Image Forgery Detection

Abstract

Image manipulation detection has emerged as a crucial task in the digital forensics and strategic intelligence domains. The easy accessibility of modern image-editing applications has made tampering more common and difficult to detect, malicious manipulations—particularly copy–move forgeries pose serious threats to the integrity of multimedia data. This paper proposes a hybrid deep-learning framework combining Discrete Cosine Transform (DCT), Scale-Invariant Feature Transform (SIFT), and a lightweight Convolutional Neural Network (CNN) classifier to effectively identify and localize forged regions in digital images. Experimental results demonstrate a detection accuracy of 96.8% under various compression and transformation scenarios, outperforming classical DCT- and SIFT-based methods.