Image Forgery Detection through Fusion of Xception CNN and Thepade SBTC Features using ML Classifiers and Ensembles

In today’s dynamic digital landscape, verifying the authenticity of visual content is paramount. This research tackles the crucial task of image forgery detection that is distinguishing genuine images from manipulated ones. We propose a novel approach by harnessing the capabilities of Xception CNN for intricate pattern recognition and Thepade SBTC features for interpretability and resilience. This study employs eight ML algorithms and top performing ensembles to assess effectiveness of our fusion methodology. Results indicate that ensembles surpass in- dividual algorithms, achieving superior image forgery detection. Specifically, fusing Thepade SBTC and Xception CNN features demonstrates remarkable performance, outperforming individual feature sets. These findings not only propel image forensics but also provide a practical solution for safeguarding digital visual integrity in today’s technology-driven milieu.