Multi-Scale FPGA-Based Infrared Image Enhancement by Using RGF and CLAHE

Area of Science:

• Image processing
• Computer vision
• Embedded systems

Background:

• Infrared sensors are vital for surveillance, autonomous driving, and medical diagnostics, operating in diverse weather conditions.
• Infrared imagery suffers from low contrast and indistinct textures, hindering visual analysis.
• Existing enhancement algorithms are computationally intensive, challenging real-time processing.

Purpose of the Study:

• To develop a multi-scale, FPGA-based method for real-time infrared image enhancement.
• To address challenges of low contrast and indistinct textures in infrared images.
• To improve the visual quality and detail of infrared imagery for practical applications.

Main Methods:

• A multi-scale decomposition of infrared images using the rolling guidance filter (RGF).
• Fusion of detail layers with gain coefficient enhancement and contrast enhancement of the base layer using CLAHE.
• Implementation of the proposed algorithm on an FPGA using high-level synthesis tools for real-time processing.

Main Results:

• The proposed method significantly improves image contrast and enhances detail information.
• Real-time enhancement achieved at 147 FPS for 640 × 480 infrared images.
• Demonstrated effectiveness on the AXU15EG board.

Conclusions:

• The developed FPGA-based method offers an effective solution for real-time infrared image enhancement.
• The combination of RGF and CLAHE successfully addresses contrast and detail issues.
• High-speed processing capability makes the method suitable for demanding applications.