Research On Lane Detection And Vehicle Identification Technology In Intelligent Vehicle Vision System

Nowadays,With the rapid development of global auto industry,the traffic safety problem caused by the driver's wrong operation is increasingly serious.The active safety technology represented by intelligent auxiliary driving system can effectively reduce the traffic accidents caused by driver's error.In order to reduce traffic accidents due to driver's inattention and strengthen the safety of vehicle,this paper design the lane and vehicle detection algorithm based on machine vision,which can not only provide lane departure warning to the driver accurately,but also can provide vehicle collision warning.A distributed lane detection algorithm based on voting mechanism and regional planning is proposed,which can be used in lane detection and lane tracking.Firstly,the region of interest is divided by the captured image,and the feature points of the lane mark are extracted by the information such as the width of lane line.Secondly,the vanishing point is located according to the position and the direction angle of the feature points,and set up relevant voting space to detect vanishing point and lane edge gradient,the target fitting area is then defined to extract the lanes.Finally,the strip-shaped region is defined to track lanes based on confidence judgement.This paper conducts a thorough research on lane departure warning model of different types,to design lane departure warning algorithm and set up warning trigger conditions.Futhermore,the key technologies of collision warning system are also studied to design a vehicle identification algorithm based on machine learning.The positive and negative training samples of vehicle are collected,the samples are trained by the adaboost algorithm based on the Haar-like feature to acquire the cascade classifier.The model of distance measurement based on monocular vision is also studied.The car test using IMX6 S as lane departure warning test platform to verify the effectiveness of lane detection algorithm.the experimental results show that the algorithm has good real-time performance and accuracy in the embedded system.In addition,the fusion test of vehicle and lane detection algorithm is carried out,the proposed algorithm can accurately detect vehicles and lanes,and basically meets the real-time and accuracy requirements of the system.