Generalizable AI-Generated Image Detection Based on Fractal Self-Similarity in the Spectrum

2025-03-11

AI Summary

AI-generated image detection suffers from poor generalization to unseen generative models. To address this, we propose a universal detection method grounded in spectral fractal self-similarity. We are the first to identify and model a cross-model-consistent fractal self-similar structure inherent in the frequency spectra of AI-generated images, replacing conventional spectral analysis with a similarity metric between fractal branches. Our approach integrates spectral analysis, fractal geometric modeling, periodic extension, low-pass filtering, and cross-model feature consistency learning. Evaluated on public benchmarks, the method achieves significantly higher detection accuracy for images synthesized by unseen GANs and diffusion models compared to state-of-the-art techniques. It demonstrates strong cross-model generalization—without requiring retraining or fine-tuning on target models—thereby establishing a new paradigm for universal AI-generated image detection.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

The generalization performance of AI-generated image detection remains a critical challenge. Although most existing methods perform well in detecting images from generative models included in the training set, their accuracy drops significantly when faced with images from unseen generators. To address this limitation, we propose a novel detection method based on the fractal self-similarity of the spectrum, a common feature among images generated by different models. Specifically, we demonstrate that AI-generated images exhibit fractal-like spectral growth through periodic extension and low-pass filtering. This observation motivates us to exploit the similarity among different fractal branches of the spectrum. Instead of directly analyzing the spectrum, our method mitigates the impact of varying spectral characteristics across different generators, improving detection performance for images from unseen models. Experiments on a public benchmark demonstrated the generalized detection performance across both GANs and diffusion models.

Problem

• Improves detection of AI-generated images from unseen models
• Exploits fractal self-similarity in image spectra for generalization
• Addresses accuracy drop in detecting images from unknown generators

Innovation

• Fractal self-similarity in spectrum analysis
• Periodic extension and low-pass filtering
• Generalized detection across GANs and diffusion models