基于计算机视觉和霍夫变换的车道线检测研究  

Research on lane detection based on computer vision and Hough transform

作　　者：刘月 杨伟樱 Liu Yue;Yang Weiying(Shaanxi Industrial Vocational and Technical College,Information Engineering College,Xianyang,Shaanxi 712000)

机构地区：[1]陕西工业职业技术学院,信息工程学院,陕西咸阳712000

出　　处：《现代科学仪器》2023年第2期144-148,共5页Modern Scientific Instruments

基　　金：陕西工业职业技术学院自然科学研究项目(编号:2022YKYB-013)。

摘　　要：为提升车道线的检测准确率,本文提出了一种基于计算机视觉和霍夫变换的高速公路车道线检测新方法。为解决弯曲车道线的检测问题,对车道线图像按区段划分,并通过直线车道线来逼近每一小段弯曲车道线。然后,对直线车道进行霍夫变换,提取直线车道特征。最后,将检测到的短直线车道的端点作为输入,通过插值算法来近似车道线。对不同状况实际车道线的测试结果表明,所提出的检测算能够以可接受的像素偏差实现左右车道线的准确检测。To improve the accuracy of lane line detection,this paper proposes a new expressway lane line detection method based on computer vision and Hough transform technology.In order to solve the detection problem of curved lane lines,the lane line image is divided into sections.Then each small curved lane line is approached using a straight lane line.After that,Hough transform is applied to each linear lane to extract the characteristics of this linear lane.Finally,the endpoints of the detected short straight lanes are used as input to approximate the lane lines by interpolation algorithm.The test results of actual lane lines under different conditions show that the proposed algorithm can accurately detect left and right lane lines with acceptable pixel deviation.

关 键 词：车道线检测 计算机视觉 图像区段划分 插值 霍夫变换 

分 类 号：U495[交通运输工程—交通运输规划与管理]