Lane Line Recognition Of Vehicle Under Different Illumination Conditions

In view of the current development stage of vehicle intelligence,this paper studies the lane line recognition technology of auxiliary driving system and even automatic driving vehicles.The current lane recognition algorithm is greatly affected by the illumination.So the illumination is introduced into the algorithm to study the lane line recognition algorithm under different illumination environments.The algorithm can complete the lane line recognition function at the vehicle speed of 30km/h,and the average recognition accuracy can reach 90.30%.The average frame rate of lane recognition algorithm under different illuminations can reach 402fps,which can meet the experimental requirements.Firstly,illumination is taken as an important parameter of illumination affecting image quality.So illumination classifier is designed,and the illumination collection system platform is built to complete the function of real-time collection of ambient illumination.The illumination collection system platform is composed of BH1750 illumination sensor,CH340 chip USB TO TTL and raspberry pie 4B.Secondly,optical flow estimation and background modeling is used to highlight the lane line pixels in a dynamic environment under normal illumination.For lane line recognition under low illumination and night illumination,this paper uses the statistical direction gradient histogram method to process road images.The light spot or light dark junction often appear in the strong illumination images and the lane line information is easy to be lost.So the threshold segmentation histogram equalization is used as the illumination compensation algorithm to process the images.Thirdly,hough transform is used to locate the lane line,and the least square method is used to fit the lane line.The method can be adapted to the straight and curve lane lines.Finally,in the experimental verification of the algorithm,the feasibility of the contrast degree classifier can be verified by collecting the road illumination data of different periods.At the same time,real vehicle road test is carried out in different illumination environment to verify the recognition effect.By comparing and analyzing the recognition effect,efficiency and accuracy,the lane line recognition algorithm under different illumination can adapt to the road image under different illumination.The recognition rate and accuracy can meet the experimental requirements.It is significant to the research of intelligent vehicle lane line recognition based on vision.