基于机器视觉的红外激光辅助车道线检测方法  

Infrared laser-assisted lane line detection method based on machine vision

作　　者：李伟林 方遒[1,2] LI Weilin;FANG Qiu(Xiamen University of Technology,Xiamen Fujian 361024,China;Xiamen University,Xiamen Fujian 361005,China)

机构地区：[1]厦门理工学院,福建厦门361024 [2]厦门大学,福建厦门361005

出　　处：《激光杂志》2023年第5期58-65,共8页Laser Journal

基　　金：福建省科技计划项目(No.2021J01230568)。

摘　　要：车道线检测在车道偏离预警、车道保持、车道变换、前向碰撞预警、自适应巡航控制等驾驶辅助系统中发挥着重要的作用。传统的车道线检测算法存在算法复杂度较高、在路面变化时识别率低等问题。针对以上问题,提出了一种红外激光辅助车道线检测的方法。通过两道激光照射路面后激光光条在图像中的位置获取路面角度信息,以实现自适应逆透视变换;使用Canny算法完成二值化、Hough变换检测直线;利用激光照射位置处路面和车道线灰度差值明显的特点定位车道线,完成直线的筛选。实际路况的实验表明,所提算法能够有效地应对路面变换带来的视角变化并减少环境干扰,相比固定ROI提取方法和传统Hough直线检测方法效果更佳、鲁棒性更强,平均正确率达到97.24%,平均运算时间为28.7 ms/帧。Lane line detection play an important role in driving assistance systems such as lane departure warning,lane keeping,lane change,forward collision warning,and adaptive cruise control.The traditional lane line detection algorithm has the problems of high algorithm complexity and low lane line recognition rate when the road surface changes.Aiming at the problems above,an infrared laser-assisted lane line detection method is proposed.The position of the laser light bar in the image after two lasers irradiate the road surface is used to obtain the road surface angle information to achieve adaptive reverse perspective transformation.Use Cranny’s algorithm to complete binarization and Hough transform to detect straight lines.The lane line is located by using the obvious difference between the gray level of the road surface and the lane line at the laser irradiation position,and the straight line screening is completed.Experiments on actual road conditions show that the proposed algorithm can effectively deal with the change of perspective caused by road surface changes and reduce environmental interference.Compared with the fixed ROI extraction method and the traditional Hough line detection method,the effect is better,the robustness is stronger,and the average accuracy rate reaches 97.24%,and the average operation time is 28.7 ms/frame.

关 键 词：自适应 逆透视变换 车道线检测 图像预处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] U491.222[自动化与计算机技术—计算机科学与技术]