Simulation Of Lane Recognition Based On ADAS Experimental Platform

The progress of the times drives the rapid development of the automobile industry.Automobile safety has become a hot topic.Assisted driving is the focus of the research in the field of automobile safety.Lane line recognition is the basis for most functions of auxiliary driving.Because of the limited conditions of auxiliary driving experiments,it is necessary to simulate the lane line identification.This paper mainly studies the lane line recognition function of ADAS experimental platform and makes simulation experiment verification.(1)In the image preprocessing part,the interested areas are divided firstly,and one second of the area under the image is taken as the interested area;the principle and difference of different grayscale methods are analyzed,and the weighted average method is selected according to the characteristics of the collected data to gray the image;the common filtering calculation principle,algorithm template selection and algorithm application range are introduced.According to the noise situation of the image,high-level image noise is adopted In order to identify the contour of lane line,the binary operation is carried out for the image;finally,the edge extraction operation is carried out,and the advantages and disadvantages of different operators are compared,and Canny operator with better extraction effect is selected for processing.(2)In the lane line identification section,the limit conditions of lane line are set according to national road standards,and the virtual lane line model is constructed by combining parametric method with the restriction condition.In order to improve Hoff algorithm,only one lane line can be identified,extreme radius polar angle constraint is added to traditional Hoff transform method to realize accurate recognition of two lane lines;the least square lane line is introduced The method mainly limits the search of feature points,including the gradient amplitude,direction and geometric parameters of lane.Finally,the advantages and disadvantages of the two methods are analyzed.Simulation results show that both algorithms can identify lane lines accurately,and both of them realize the lane line recognition function of the platform.(3)In the image defogging part,the causes of fog and the characteristics of fog images are analyzed,and the mainstream defogging algorithms are sorted out and typical algorithms are introduced.The paper introduces the model of ambient light and atmospheric propagation,introduces the dark primary color defogging algorithm based on the two models.Aiming at the rough value of the calculation of the transmittance,the abandonment factor is introduced to improve the accuracy of the transmittance estimation and the tolerance is introduced to solve the problem of poor defogging effect in the large area of highlighted area.The simulation results show that the improved algorithm can effectively accomplish the fog removal of virtual lane line.(4)In the simulation experiment part,the software and hardware of the experimental platform are introduced firstly,including single vision acquisition platform,Ni operation platform and simulation software prescan,Car Sim and MATLAB.Then,the parametric method is used to generate virtual scenes of guardrail Road,shadow Road,fog Road,multi lane and lane deviation by using prescan software.The fog removal experiments were carried out in different scenes,and the lane line recognition algorithm was verified in the lane departure scene by multi software combined simulation.The experimental results show that the image defogging algorithm is effective and the goal of defogging is completed.The simulation method of lane line recognition can achieve the predetermined target,which lays a foundation for the subsequent development of machine vision.