Research On Key Algorithms Of Lane Departure Warning System Based On Machine Vision

With cars becoming an indispensable means of transportation in our daily life,road traffic accidents are becoming a more and more serious social problem.LDWS can give early warning information when a vehicle deviate from the normal lane,which can greatly reduce the occurrence of serious traffic accidents.Lane line detection algorithm and lane departure warning algorithm are the most critical algorithms in LDWS.The main research contents of this paper are as follows:First of all,in the image preprocessing step of the image collected by the on-board camera,the weighted average method is used to grayscale the image,in order to reduce the processing amount of image data,a method to select the ROI for fixed target detection distance is proposed,median filter is used to reduce image noise.The problem of low detection accuracy is caused by the fact that lane line feature extraction is easily affected by complex light.The illumination classification is designed for the collected images,The linear enhancement algorithm and the histogram tapered stretch algorithm are used to enhance the images under low and strong illumination conditions.Secondly,in the lane line detection phase,a new threshold segmentation algorithm is proposed in the stage of image binarization The global threshold and the local reference threshold value are combined with the new threshold value to perform binarization processing on the image,which can effectively improve the adaptability of the algorithm to the complex lighting environment.Then the Canny operator was used to extract the edge features of the lane line,to solve the problem that the high and low threshold in the Canny operator was not easy to determine,the iterative method was used to determine the high and low threshold in the Canny operator.Finally,the improved Hough transform is used to fit the lane line.Then,in the lane departure warning of judgment,a reasonable tracking strategy based on Kalman filter is designed and combined with the actual situation,this paper put forward a kind of deviating from the early warning of judgment based on the vehicle lateral distance and lane line angle,improve the early warning and accuracy.Finally,three data sets containing different lighting and road scenes are used to test the lane line algorithm.The experimental results showed that the comprehensive recognition rate of lane lines under different lighting and road environments was 95.61%,and the average detection time of single frame was 34.51 ms.An experimental study was carried out on lane deviation early warning including 2200 frames of road driving video.The accuracy rate of lane deviation early warning was 94.92%.The experimental results show that the algorithm in this paper can quickly and accurately identify lane lines and realize deviation warning.This paper studies two key algorithms in LDWS,lane line detection and lane departure warning algorithm,and provides ideas for intelligent vehicles to study LDWS.