基于多约束条件下的霍夫变换车道线检测方法  

Hough Transform Lane Detection Based on Multiple Constraints

作　　者：石林军 余粟 Shi Linjun, Yu Su(Engineering Practice and Training Center , Shanghai University of Engineering Science, Shanghai 201620, Chin)

机构地区：[1]上海工程技术大学工程实训中心,上海201620

出　　处：《计算机测量与控制》2018年第9期9-12,38,共5页Computer Measurement &Control

基　　金：国家科技支撑计划(2015BAF10B00)

摘　　要：车道线检测应用越来越实际,对车道线提取要求也越来越高,为了克服复杂多变的车道环境,提出一种通用环境的车道线提取方法;从原始RGB图像的每个颜色通道获取信息开始,获得鲁棒性很好的灰度化图像;采用一种多约束下的Hough变换提取特征线,接着在提取的线段基础上用概率表决程序估计消失点,接着用消失点约束车道线候选线;最后对剩下的特征线K-mean聚类;试验结果表明,该方法提取车道线鲁棒性很好,检测精度高,识别率97%以上,并处理时间较短,实时性好。The application of lane line detection is more and more practical,and the requirement of lane line extraction is also getting higher and higher.In order to overcome the complex and changeable lane environment,ageneral environment lane line extraction method is proposed.Get raw information from each color channel in your original RGB image and get a very robust grayscale image.A multi-constrained Hough transform is used to extract the feature line,and then the vanishing point is estimated based on the extracted line segment using the probability voting procedure.Then,the vanishing point is used to constrain the lane line candidate segment.Finally,the rest of the characteristic line K-mean clustering.The experimental results show that the proposed method has good robustness,high detection accuracy,recognition rate more than 97%,short processing time and good real-time performance.

关 键 词：ROI HOUGH变换 消失点估计 K-mean聚类 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]