Color Matters: Demosaicing-Guided Color Correlation Training for Generalizable AI-Generated Image Detection

2026-01-30

AI Summary

This work addresses the limited generalization of existing AI-generated image detectors to unseen generative models by proposing a self-supervised U-Net approach that models the conditional inter-channel dependencies introduced by demosaicing in real camera imaging. Specifically, it simulates the color filter array (CFA) and demosaicing pipeline to capture intrinsic color correlations inherent in authentic photographs. For the first time, demosaicing-guided color correlation modeling is introduced into the detection of synthetic images, revealing provable distributional discrepancies in color features between real and generated images. The method achieves state-of-the-art cross-model generalization and robustness across more than 20 unseen AI generators, significantly outperforming current techniques.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• 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

Abstract

As realistic AI-generated images threaten digital authenticity, we address the generalization failure of generative artifact-based detectors by exploiting the intrinsic properties of the camera imaging pipeline. Concretely, we investigate color correlations induced by the color filter array (CFA) and demosaicing, and propose a Demosaicing-guided Color Correlation Training (DCCT) framework for AI-generated image detection. By simulating the CFA sampling pattern, we decompose each color image into a single-channel input (as the condition) and the remaining two channels as the ground-truth targets (for prediction). A self-supervised U-Net is trained to model the conditional distribution of the missing channels from the given one, parameterized via a mixture of logistic functions. Our theoretical analysis reveals that DCCT targets a provable distributional difference in color-correlation features between photographic and AI-generated images. By leveraging these distinct features to construct a binary classifier, DCCT achieves state-of-the-art generalization and robustness, significantly outperforming prior methods across over 20 unseen generators.

Problem

• AI-generated image detection
• generalization
• color correlation
• demosaicing
• camera imaging pipeline

Innovation

• Demosaicing
• Color Correlation
• AI-Generated Image Detection
• Self-supervised Learning
• Camera Imaging Pipeline