Research On Lane Detection Method Based On A Hybrid Model In Urban Environment

At present,traffic accidents occur frequently,and intelligent driving technology has become one of the effective means to solve this problem.Lane detection is the basic and important part of intelligent driving technology,and it plays an important role in vehicle path planning and decision-making.To the best of our knowledge,it is difficult to extract the lane target information based on vision in the urban environment,and the accuracy of detection is challenging.This thesis studies the problem of lane detection using the vehicle camera.In order to solve this problem,the feature of lane detection is extracted.Then,a contrast stretching algorithm based on multiple threshold segmentation and a pretreatment method of zebra crossing filtering are proposed to remove the outlier from the illumination and urban road environmental interference.Furthermore,a lane detection and tracking algorithm based on a hybrid model is developed to improve the accuracy,robustness,and efficiency of lane detection.The main contributions of this thesis are given as follows:1.A contrast stretching algorithm and a pretreatment method of zebra crossing filtering based on multiple threshold segmentation are proposed.Firstly,in view of the problems of illumination and road environment interference in urban environment,a multiple threshold segmentation contrast stretching algorithm based on the largest inter-class variance is proposed.The experiment shows that the algorithms proposed for the target characteristics in this thesis are with a better adaptability and anti-interference ability than traditional pretreatment methods in urban environment.2.A lane detection method based on a hybrid model and a lane tracking method are proposed.In order to further improve the accuracy,robustness and efficiency of lane detection,considering that the single model is not greatly adaptable to road lane detection,a straightcurve hybrid model is proposed to divide the road into near and far field of vision.Firstly,in the near field,a hough straight line detection algorithm based on constraint conditions is constructed.Then,in the far field,an algorithm based on dynamic sliding window search for extracting curve feature points is proposed.Finally,a random sampling consistency algorithm for curve fitting is given.Therefore,a hybrid model based on straight lines and curves is constructed,and lane tracking is performed to further improve the robustness of lane detection.3.A real car experiment platform is built.The performance of the proposed method is verified by simulation and real intelligent vehicle experiments.The experimental results show that the algorithm proposed in this thesis has better accuracy,robustness and efficiency of detection in urban complex environments.In general,the detection method proposed in this thesis is applied to complex urban road environments,and it has certain theoretical significance and practical value for the future research on lane detection of urban roads.