Single Image Haze Removal from Image Enhancement Perspective for Real-Time Vision-Based Systems

Area of Science:

• Computer Vision
• Image Processing
• Hardware Acceleration

Background:

• Outdoor vision systems are degraded by atmospheric turbidity (haze).
• Haze removal algorithms are crucial pre-processing steps, but efficient, fast methods are needed.
• Existing methods often rely on complex physical models, limiting real-time application.

Purpose of the Study:

• To propose an efficient single image haze removal algorithm.
• To develop a hardware implementation for real-time processing of hazy images.
• To offer a computationally inexpensive yet effective solution for outdoor vision challenges.

Main Methods:

• Exploits computationally efficient image processing techniques: detail enhancement, multiple-exposure image fusion, and adaptive tone remapping.
• Avoids inverting the physical model of haze formation.
• Features a compact hardware implementation using a Field Programmable Gate Array (FPGA).

Main Results:

• Achieves good performance compared to state-of-the-art methods.
• Demonstrates low computational complexity.
• Enables real-time processing at rates exceeding 25 frames per second.

Conclusions:

• The proposed algorithm offers an efficient and fast solution for single image haze removal.
• The hardware implementation facilitates real-time applications for outdoor vision systems.
• Software and datasets are publicly available, promoting further research and development.