基于激光雷达点云的目标识别与跟踪方法  

Object Classification and Tracking Approach Using LiDAR Point Clouds

作　　者：郭子明 GUO Ziming(Changzhou Xingyu Automotive Lighting Systems Co.,Ltd.,Changzhou 213001,China)

机构地区：[1]常州星宇车灯股份有限公司,江苏常州213001

出　　处：《照明工程学报》2023年第5期64-67,共4页China Illuminating Engineering Journal

摘　　要：对复杂驾驶环境中移动目标的准确识别与跟踪是驾驶辅助系统需要解决的重要问题之一。本文提出了一种基于激光雷达点云的目标识别与跟踪方法,首先对每一帧点云数据进行聚类并提取目标线段,利用采集的大量实验数据获取各类目标的特征值参考区间,然后通过特征匹配的方式完成目标类别的识别,最后基于扩展卡尔曼滤波(EKF)对不同类别的目标状态进行预测与更新。实验结果表明,相较于既有的基于特征匹配的目标识别方法,本文的方法能够明显提升目标识别性能,并能实现稳定的目标跟踪。One of the most important problems to be addressed for the driving assistance systems is the accurate classification and tracking of moving objects in the complex driving environment.This paper presented a object classification and tracking approach using LiDAR point clouds.Each frame of the point clouds was grouped and the object line segments were extracted.The feature reference intervals of different classed of the objects were generated based on the large amount of experiment data.Then the object classification was achieved using feature matching.The states of different objects were predicted and updated within the EKF process.The experimental results show that the approach proposed can improve the performance of object classification remarkably and achieve stable object tracking.

关 键 词：特征提取 目标识别 目标跟踪 扩展卡尔曼滤波 

分 类 号：TM923[电气工程—电力电子与电力传动]