基于多特征融合和窗口搜索的新型车道线检测算法  

Novel lane detection algorithm based on multi-feature fusion and windows searching

作　　者：蔡创新 邹宇[1] 潘志庚 刘志彬 高尚兵 CAI Chuangxin;ZOU Yu;PAN Zhigeng;LIU Zhibin;GAO Shangbing(School of Artificial Intelligence,Nanjing University of Information Science and Technology,Nanjing,Jiangsu 210044,China;Faculty of Computer and Software Engineering,Huaiyin Institute of Technology,Huaian,Jiangsu 223001,China)

机构地区：[1]南京信息工程大学人工智能学院,江苏南京210044 [2]淮阴工学院计算机与软件工程学院,江苏淮安223001

出　　处：《江苏大学学报（自然科学版）》2023年第4期386-391,共6页Journal of Jiangsu University：Natural Science Edition

基　　金：国家自然科学基金资助项目(62072150);馆藏资源活化技术文化和旅游部重点实验室开放课题项目(CRRT2022K03);浙江省自然科学基金资助项目(LZ21F020008);江苏省“333工程”项目(BRA2016454);江苏省教育科学“十四五”规划2021年度课题项目(C-c/2021/03/30);2022年度高校哲学社会科学研究一般项目(2022SJYB0852)。

摘　　要：针对现有车道检测算法准确性和实时性较难平衡的问题,提出了一种基于多特征融合和窗口搜索的新型车道线检测算法.采用多边形填充方法确定车道线的感兴趣区域(region of interest, ROI),融合车道线的颜色、直方图和梯度特征,以消除ROI中的复杂背景.通过单应性变换得到车道线的二值图像,基于其像素密度分布寻找车道线初始位置,以窗口搜索方式提取整个车道线上的所有候选像素点.通过拟合像素点构建车道线数学模型.结果表明:提出的算法具有较高的准确性和实时性,算法对黄色车道线、树木阴影遮挡、光照变化、车道线缺损和地面交通标志干扰具有较好的鲁棒性.To solve the problem that the existing lane detection algorithms were difficult to balance accuracy and real-time performance,a novel lane detection algorithm was proposed based on multi-feature fusion and windows searching.The polygon filling method was used to determine the region of interest(ROI)of the lane lines,and the color,histogram and gradient features of the lane lines were fused to eliminate the complex background in the ROI.The binary image of the lane lines was obtained by homography transformation,and the initial position of the lane lines was determined based on the pixel density distribution.The entire candidate pixel points along the lane line were extracted by the windowbased searching method,and a mathematical model of the lane line was constructed by fitting the extracted pixels.The results show that the proposed algorithm has high accuracy and real-time performance,and the algorithm has excellent robustness to various factors of yellow lane lines,tree shade obstruction,illumination changes,missing lane lines and interference from ground traffic signs.

关 键 词：无人驾驶 车道线检测 多特征融合 单应性变换 窗口搜索 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]