毫米波雷达与视觉传感器信息融合的车辆跟踪  

Vehicle Tracking of Information Fusion for Millimeter-wave Radar and Vision Sensor

作　　者：胡延平[1] 刘菲 魏振亚[2] 赵林峰[2] HU Yanping;LIU Fei;WEI Zhenya;ZHAO Linfeng(School of Mechanical Engineering,Hefei University of Technology,Hefei,230009;School of Automotive and Traffic Engineering,Hefei University Technology,Hefei,230009)

机构地区：[1]合肥工业大学机械工程学院,合肥230009 [2]合肥工业大学汽车与交通学院,合肥230009

出　　处：《中国机械工程》2021年第18期2181-2188,共8页China Mechanical Engineering

基　　金：国家自然科学基金(51675151,U1564201);安徽省科技重大专项(17030901060);汽车新技术安徽省工程技术研究中心开放基金(QCKJ202002)。

摘　　要：为提高车辆前向防碰撞预警系统对前方道路环境感知的准确性,提出一种毫米波雷达与视觉传感器信息融合的车辆跟踪方法。提出的剔除雷达干扰目标算法缩短了对干扰目标的处理时间;提出的对称检测算法可对雷达目标兴趣区域进行对称检测,减小雷达目标兴趣区域的横向位置误差;提出的KCF-KF组合滤波算法可提高跟踪车辆的精度。实车试验表明,该方法可有效跟踪车辆位置信息,像素坐标系下的X、Y坐标跟踪准确率分别超过97.34%与95.19%。In order to improve accuracy of vehicle forward collision prevention warning system on road environment perception,a vehicle tracking method of information fusion for millimeter-wave radar and vision sensor was proposed.An algorithm to eliminate radar jamming targets was proposed to reduce processing time of jamming targets.A symmetric detection algorithm was proposed to detect radar target ROI(region of interest)symmetrically and reduce lateral position errors of radar target ROI.In order to improve the tracking accuracy,a KCF-KF combined filtering algorithm was proposed to track and fuse vehicles.Actual vehicle tests show that the method may effectively track vehicle position information,and the tracking accuracy of X and Y coordinates in pixel coordinate system is more than 97.34%and 95.19%respectively.

关 键 词：毫米波雷达 视觉传感器 信息融合 车辆跟踪 

分 类 号：U121[交通运输工程]