🤖 AI Summary
This study addresses the high computational cost of existing image forgery detection methods and their susceptibility to format-induced confounding bias, which often leads models to rely on file format artifacts rather than genuine manipulation traces. To overcome these limitations, the authors propose a lightweight, interpretable feature engineering pipeline that enables efficient image-level forgery screening using only CPU resources and gradient-boosted trees. The approach constructs a 405-dimensional feature vector by integrating multi-scale Error Level Analysis (ELA) across seven JPEG quality levels, cross-quality ELA ratios, and additional handcrafted features, effectively capturing inconsistencies in compression history while mitigating format confounding. Evaluated on the CASIA v2.0 JPEG subset, the model achieves an AUC of 0.990 (95% CI: 0.988–0.991) and an F1 score of 0.905; under source-aware partitioning, it maintains strong performance with an AUC of 0.976, and processes each image in under one second.
📝 Abstract
Image forgery detection is a critical task in digital forensics, yet many deep-learning localization approaches are typically GPU-accelerated and computationally heavier than handcrafted screening methods. We propose a lightweight, interpretable feature engineering pipeline for image-level forgery screening using only CPU computation and gradient boosted trees. Our method introduces \emph{multi-scale Error Level Analysis} (ELA) computed at seven JPEG quality levels, combined with novel \emph{cross-quality ELA ratio} features that capture double-compression artifacts characteristic of spliced regions, augmented by spatial entropy, FFT energy bands, edge density, SRM residuals, and DCT blockiness, yielding a 405-dimensional feature vector. CASIA v2.0 contains a format confound (60\% of tampered images are TIFF while authentic images are JPEG/BMP and contain no TIFF samples), enabling a trivial \texttt{is\_tiff} classifier to reach 0.80 AUC. We address this through rigorous format-controlled evaluation: on the JPEG-only subset (9,501 images, eliminating the TIFF/JPEG container confound), our method achieves AUC~=~0.990 [95\% CI: 0.988--0.991] and F1~=~0.905 using 5-fold stratified cross-validation. Under a conservative source-aware group split (preventing related images from appearing in both train and test), AUC remains 0.976. An ablation study reveals that multi-scale ELA provides the dominant gain (+0.180 AUC over single-quality on the format-controlled subset), while cross-quality ratios provide complementary double-compression detection. These results support that the method detects compression-history inconsistencies rather than file-format shortcuts -- while offering feature-level interpretability, CPU-only deployment, and sub-second inference.