Lane Detection And Departure Warning Technology Based On Smartphone

As the main means of transportation,cars make people travel comfortably and conveniently,but also bring about traffic safety hazards.In the process of driving,due to the limited visibility of the driver,the complicated traffic environment,and the driver’s rusty or fatigued state,the accident caused by the vehicle deviates from the driving lane happened in the driver’s unconsciousness.Not only does it cause loss of time and property,but also leads to traffic jam in some areas,it may cause serious damage to personal safety.As the key technology of active safety,Lane Departure Warning Systems(LDWS)avoids accidents by means of alarm driver.Therefore,LDWS is of great significance for road traffic safety.This thesis aims at the complex application process of the existing LDWS,the relatively large number of components that need to be installed,and the relatively high cost.With the current popularity of the Android mobile terminal and the processing performance has been greatly improved,this thesis mainly studies the lane detection and departure decision technology based on smartphone.The research reduces the application cost of LDWS and reduces the installation components,which lay the foundation for the popularity of LDWS.The thesis starts from the following two aspects:Firstly,due to the complexity and variability of the actual traffic environment,all the light changes,shadow occlusion and road surface damage have certain impacts on lane detection.Based on this,this thesis proposes a lane detection algorithm in complex environment.In the lane detection stage,the algorithm uses Top-Hat to enhance the gray contrast between lanes and background in the preprocessing image.Then we combine the adaptive Threshold and edge detection to obtain the candidate feature points of lanes.Meanwhile,according to the structural characteristics of actual lanes,the Hough space is constrained to fit lanes and lanes on both sides of the vehicle are tracked by Kalman.The data set collected in Xi’an and the data set shared on the network are simulated and analyzed finally.It is verified that the algorithm can adapt to different scenarios and has a higher accuracy rate under the condition of meeting real-time requirement.Secondly,in the actual driving scenario,the traditional CCP model cannot meet the demand of vehicles driving near the lane boundary.Therefore,based on actual demand,this thesis first uses the relative lateral offset of the vehicle and the direction angle of lanes to determine whether the vehicle is into departure status,then combines the change trend of relative lateral offset to estimate vehicle movement,and finally determines whether the vehicle is about to deviate from the current lane.Experimental results show that the departure decision algorithm based on vehicle movement can improve the warning accuracy rate.Finally,according to the current application status of the LDWS,this thesis analyzes the problems that the current system is not widely used.Based on the Android mobile terminal,this thesis develops a lane departure warning application,and verifies the effectiveness of the above algorithm,so as to provide a lane departure warning service with low-complexity,low-cost,easy-to-install.