CDFFusion: A Color-Deviation-Free Fusion Network for Nighttime Infrared and Visible Images

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Infrared and visible image fusion aims to combine complementary information for enhanced image representation.
• Existing fusion methods struggle with low-light conditions, leading to issues like overexposure, blocking, and color deviation.

Purpose of the Study:

• To propose CDFFusion, a novel image fusion method specifically designed for low-light scenarios.
• To address the limitations of existing methods in extracting dark details and preventing visual artifacts.

Main Methods:

• Utilizes Retinex theory to decompose visible light images into illumination and reflection components.
• Fuses reflection features with infrared features to generate the Y component of the fused image.
• Applies a color mapping formula to adjust Cb and Cr components, then concatenates for the final fused image.

Main Results:

• Achieved superior performance on the LLVIP dataset with metrics like SF (17.6531) and MS-SSIM (0.8335).
• Effectively alleviates visual overexposure and image blocking effects.
• Demonstrates the smallest color deviation among evaluated methods.

Conclusions:

• CDFFusion offers an effective solution for infrared and visible image fusion in low-light environments.
• The method successfully enhances information expression while preserving image fidelity and color accuracy.