Lane Line Recognition Based On Image Processing

In increasing the number of cars lead to traffic jams on the road,increasing the chance of traffic accidents,the lane departure warning system plays a big role in reducing traffic accidents and safe driving.An important part of the lane departure warning system is lane line detection and tracking,which can accurately and effectively identify lane lines,so it is of great significance to study lane line identification.The paper firstly preprocesses the original road image for the problem that the actual lane line feature information is not obvious,including lane line color segmentation,weighted average grayscale road image,median filtering to eliminate image noise,and OTSU method binarized image.Sobel operator detects the edge of the lane line and selects the region of interest from the road image.Based on the above methods,the irrelevant information in the image can be eliminated,and the characteristics of the detected lane line can be enhanced.the detected lane line features are enhanced.Secondly,for the problem of false detection and missed detection,according to the analysis of road type and lane line detection model,structured road and straight line and hyperbolic model are selected,the assumptions of road type and lane line detection model are set by actual conditions.The lane line information detected by the Hough transform is obtained by using the sliding window.The constructed line and curve model are used to match the obtained lane line information to determine the model parameters,and then the lane line position is accurately detected.Finally,there may be deviations in the detection results,and the lane line is tracked by Kalman filtering to ensure the validity of the lane line detection results.The simulation proves that the proposed method can accurately and effectively identify the lane lines in the actual road.