Enhancing Frequency Forgery Clues for Diffusion-Generated Image Detection

2025-11-01

AI Summary

Diffusion-based image detectors suffer from poor generalization and weak robustness, particularly against unseen diffusion models and post-processing/compression distortions. Method: We observe that diffusion-generated artifacts exhibit progressive frequency-domain discrepancies—from low to high frequencies—and propose a novel frequency-selective Fourier spectrum method. This approach employs an adaptive frequency selection function to weight the Fourier spectrum, enhancing discriminative frequency bands while suppressing redundant information, thereby amplifying forensic cues across the full frequency spectrum. Contribution/Results: The method is model-agnostic and distribution-agnostic, requiring no access to specific diffusion architectures or training data distributions. It significantly improves generalization to unseen diffusion models and robustness to diverse perturbations (e.g., JPEG compression, resizing, blurring). Evaluated on multiple benchmark datasets, it outperforms existing state-of-the-art detectors, demonstrating the effectiveness and universality of frequency-domain modeling for generalized generative image forensics.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Diffusion models have achieved remarkable success in image synthesis, but the generated high-quality images raise concerns about potential malicious use. Existing detectors often struggle to capture discriminative clues across different models and settings, limiting their generalization to unseen diffusion models and robustness to various perturbations. To address this issue, we observe that diffusion-generated images exhibit progressively larger differences from natural real images across low- to high-frequency bands. Based on this insight, we propose a simple yet effective representation by enhancing the Frequency Forgery Clue (F^2C) across all frequency bands. Specifically, we introduce a frequency-selective function which serves as a weighted filter to the Fourier spectrum, suppressing less discriminative bands while enhancing more informative ones. This approach, grounded in a comprehensive analysis of frequency-based differences between natural real and diffusion-generated images, enables general detection of images from unseen diffusion models and provides robust resilience to various perturbations. Extensive experiments on various diffusion-generated image datasets demonstrate that our method outperforms state-of-the-art detectors with superior generalization and robustness.

Problem

• Detecting diffusion-generated images to prevent malicious use
• Improving generalization across unseen diffusion models and settings
• Enhancing robustness against various image perturbations

Innovation

• Enhancing frequency forgery clues across all bands
• Using frequency-selective function as weighted filter
• Suppressing less discriminative bands while enhancing informative ones