STAFuse: Scene-Text Aggregation Guided Composite Degradation-Robust Infrared and Visible Image Fusion

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Infrared and visible image fusion integrates complementary data for high-quality outputs.
• Challenges include scene content differences, unpredictable degradations, and complex composite effects.
• Existing fusion models struggle with degradation robustness.

Purpose of the Study:

• Propose STAFuse, a framework for composite degradation-robust image fusion.
• Enhance fusion quality by addressing scene disparities and degradation complexities.
• Improve robustness and flexibility of image fusion models.

Main Methods:

• Developed a scene- and degradation-aware mechanism for adaptive degradation elimination.
• Utilized aggregated textual prior guidance to simulate ideal scenes for fusion.
• Implemented a textual-domain supervision loss for semantic guidance and fidelity improvement.

Main Results:

• STAFuse demonstrates significant outperformance over state-of-the-art methods.
• The framework shows enhanced robustness and flexibility in image fusion.
• Improved aggregation of complementary information and visual fidelity in fusion results.

Conclusions:

• STAFuse effectively overcomes challenges in composite degradation-robust image fusion.
• The proposed adaptive textual guidance and supervision enhance fusion performance.
• STAFuse offers a flexible and robust solution for infrared and visible image fusion.