Spectral Forensics of Diffusion Attention Graphs for Copy-Move Forgery Detection

TL;DR

GraphSpecForge is a training-free method that detects copy-move image forgeries by analyzing spectral changes in attention graphs from pretrained diffusion models, without needing forgery-specific retraining.

Contribution

It introduces a novel spectral analysis approach on attention graphs to identify copy-move forgeries, demonstrating effectiveness across multiple benchmarks.

Findings

Achieves AUROC up to 0.774 on MICC-F220 and 0.833 AUPRC on CoMoFoD.

Normalized Laplacian spectra outperform raw attention spectra by +0.057 AUROC.

Spectral signals are specific and sensitive to manipulation strength.

Abstract

Copy-move forgery, where a region within an image is duplicated to hide or fabricate content, remains a persistent threat to visual media integrity. We introduce GraphSpecForge, a training-free framework that detects copy-move forgery by analysing the spectral structure of attention graphs from a pretrained Stable Diffusion U-Net. Our central insight is that copy-move manipulation induces approximate subgraph duplication in the self-attention graph, leading to measurable spectral redistribution in the normalized graph Laplacian. We formalise this link with perturbation-based arguments and build an image-level anomaly detector using Wasserstein distances between per-image Laplacian spectra and an authentic reference distribution. We evaluate GraphSpecForge on four copy-move benchmarks without forgery-specific retraining. On RecodAI-LUC (5,128 images), our best configuration achieves AUROC = 0.606 (95% CI: 0.580-0.638; permutation p = 0.005), and the normalized Laplacian outperforms raw attention spectra by +0.057 AUROC. On MICC-F220, CoMoFoD, and COVERAGE, the same pipeline attains AUROCs of 0.752, 0.774, and 0.673, respectively; on CoMoFoD it also reaches AUPRC = 0.833, balanced accuracy = 0.712, MCC = 0.499, and TPR@1%FPR = 32.5%. Additional ablation and falsification experiments confirm the signal's specificity and sensitivity to manipulation strength, while null-graph controls rule out trivial-statistic explanations.