Lane detection in dense fog using a polarimetric dehazing method

Area of Science:

• Computer Vision
• Image Processing
• Automotive Safety

Background:

• Lane detection is vital for advanced driver-assistance systems (ADAS).
• Dense fog significantly degrades road visibility, challenging existing lane detection algorithms.
• Robust lane detection in adverse weather is critical for automotive safety.

Purpose of the Study:

• To develop an end-to-end method for effective lane detection in dense fog conditions.
• To enhance the robustness of lane detection systems under severe visibility impairment.
• To provide a computationally efficient and adaptive solution for fog-related driving hazards.

Main Methods:

• An integrated approach combining polarimetric dehazing and lane detection was proposed.
• Dense fog images captured by a monochrome polarization camera were used.
• Airlight optimization using guided filtering was performed based on polarization information.
• Lane detection was achieved using Canny edge detection and Hough transform post-dehazing.

Main Results:

• The proposed method successfully enabled lane detection in dense fog.
• The polarimetric dehazing technique effectively improved visibility.
• The system demonstrated adaptability and computational efficiency.
• Good lane detection results were achieved, enhancing driving safety.

Conclusions:

• The integrated polarimetric dehazing and lane detection method offers a robust solution for adverse weather driving.
• This approach significantly improves lane detection performance in dense fog.
• The study provides a valuable reference for enhancing driving safety in low-visibility conditions.