Realization of Anti-glare Algorithm Based on Stagnant Water Reflection of Automobile Headlights on Rainy Day

This project proposes to develop an algorithm founded on YOLOv5 to detect water accumulation at night, to eliminate the glare that results from car headlights reflecting off of standing water while people are driving at night on rainy days. Use the histogram equalization enhancement algorithm to process the night driving image to improve the contrast between the stagnant water and the road surface. Then use the YOLOv5 target detection algorithm to accurately identify where the water is located. Through the Freescale MCU to control the high and low beam switching and angle adjustment of the automobile headlights, purpose to avoid the glare of stagnant water at night. Experimental findings indicate that the histogram equalization enhancement algorithm can increase the contrast between stagnant water and road surface by 40.35%. The precision of the YOLOv5 target detection algorithm reaches more than 80%, and the precision is sufficient for practical applications. This project has high flexibility as a whole, and at the same time improves the safety of driving on rainy days at night.