Chroma Clues: Leveraging Color Statistics to Detect Synthetic Images

2026-06-01

AI Summary

This work addresses the growing challenge posed by increasingly photorealistic generative AI images to image forensics. It proposes, for the first time, a systematic and interpretable color transformation framework that exploits statistical discrepancies in chrominance between real and synthetic images. The approach integrates six handcrafted and one task-driven learned color transformations to extract color-sensitive features at either pixel or patch levels, which are then fed into a lightweight classifier for high-accuracy detection. The method achieves an average generalization accuracy of 93.27% under various post-processing attacks, enables intuitive authenticity assessment through visualizable noise patterns, and significantly enhances source attribution across diverse generative models.

Abstract

The evolution and dissemination of AI-synthesized images is occurring at an unprecedented rate. Image generators are making rapid progress in their goal of perfectly imitating natural images, which also challenges image forensics. In this work, we exploit an underexplored cue in current generative models, namely their weakness to imitate color statistics of natural images. We first show that the LPIPS loss used for training image generators is less sensitive to chrominance than to luminance, which may lead to statistical discrepancies in the colors of synthetic images. Building on this observation, we then introduce six hand-crafted color transformations and a method to learn a task-optimized color transform to statistically expose generated images. These transformations can be used in various ways. First, we define color-sensitive features at pixel-level or patch-level. A simple, interpretable classifier achieves with these features an average generalization accuracy of 93.27% and strong robustness against six types of post-processing. Second, we demonstrate that the transformations exhibit characteristic visual noise patterns in natural and synthetic image areas, which enables an intuitive visual image evaluation. Third, we demonstrate that the transforms can enhance color patterns in generated images for improved multiclass attribution.

Problem

• synthetic image detection
• color statistics
• image forensics
• generative models
• chrominance

Innovation

• color statistics
• synthetic image detection
• chrominance discrepancy
• learnable color transform
• image forensics