An Efficient Night Traffic Light Recognition Method

摘要

Traffic lights are important signals for a vehicle driving in an urban environment. In this paper, we propose an efficient night traffic light recognition method, which combine color information and region-based pixels information for detection and recognition traffic light. The proposed method firstly do pre-processing using traditional enhancement method to enhance night traffic light image/video. Color space is converted from RGB to HSI, which is used to detect candidate region of traffic light. Then, in order to remove some noise, the morphological operation is used after getting candidate region. Last, the recognition method of traffic lights based on template match and region method is employed. Experimental results show that traffic light recognition by our method is better than other conventional methods. The proposed method also shows the effectiveness.