基于激光雷达的多帧联合目标跟踪系统

秦庆国; 许录平; 孙学荣; 阎博; 孙景荣

引用本文:	秦庆国，许录平，孙学荣，阎博，孙景荣. 基于激光雷达的多帧联合目标跟踪系统[J]. 雷达科学与技术, 2022, 20(1): 98-108.[点击复制]
	QIN Qingguo, XU Luping, SUN Xuerong, YAN Bo, SUN Jingrong. Lidar Based Multi-Scan Detection in Target Tracking System[J]. Radar Science and Technology, 2022, 20(1): 98-108.[点击复制]

【打印本页】【下载PDF全文】【查看/发表评论】【下载PDF阅读器】【关闭】

←前一篇|后一篇→

过刊浏览高级检索

本文已被：浏览 6222次下载 1023次	码上扫一扫！
分享到：微信更多字体:加大+\|默认\|缩小-
基于激光雷达的多帧联合目标跟踪系统
秦庆国，许录平，孙学荣，阎博，孙景荣
1.西安电子科技大学空间科学与技术学院，陕西西安710126;2.博洛尼亚大学电子与信息工程学院，意大利47521

摘要:

本文讨论了利用激光雷达与检测前跟踪算法进行室内行人跟踪，实现监测室内人群社交距离的探测系统。激光雷达距离分辨率高，定位误差小，目标探测时间间隔小，很适用于对轨迹不确定的行人进行跟踪。本文所用检测前跟踪算法分为三步，首先利用历史数据所建杂波图滤除固定物杂波。其次利用多帧检测算法得到时间窗内的目标轨迹段，以此滤除随机噪声产生的虚警点，提升检测效率。第三步利用轨迹关联算法，将各个时间窗内的轨迹片段相互关联，得到完整的行人轨迹，从而增强机动目标与相互临近目标的跟踪效果。实测实验表明在激光雷达跟踪多个目标时，其跟踪精度在10~15cm，多次实验均未出现目标丢失或虚假轨迹，较好地完成了室内行人的检测与跟踪。

关键词: 检测前跟踪多扩展目标跟踪杂波图轨迹关联

DOI：DOI:10.3969/j.issn.1672-2337.2022.01.015

分类号:TN953；TN958.98

基金项目:教育部重点实验室项目；陕西省自然科学基金(No.2020JQ-303)；国家自然科学基金面上项目（No.62071363）

Lidar Based Multi-Scan Detection in Target Tracking System

QIN Qingguo, XU Luping, SUN Xuerong, YAN Bo, SUN Jingrong

1. School of Aerospace Science and Technology, Xidian University，Xi’an 710126，China;2. School of Electronics and Information Engineering, University of Bologna，Cesena 47521，Italy

Abstract:

An indoor tracking system using track-before-detect framework and lidar is developed to monitor the social distance. The lidar system has higher resolution, lower positional error and short detection interval. The method used here consists of three stages. At the first stage, the clutter map is built by the history measurement. And the measurement of static objects can be filtered out by the clutter map. Secondly, the tracklet in time windows are detected by the multiple scan detection based algorithm. It is designed to decrease measurement noise and increase the detection rate. At the last stage, tracklet association is applied. The tracklets in time windows are associated together to get the whole trajectory of human target. This stage is designed to ensure the detection of maneuvering target and closely distributed tracks. Our experiment shows that, using the lidar system, the positional error in tracking multiple targets can achieve 10 to 15 centimeters. In repeated experiments, no false alarm track or mis-detection arise. The experiment proves the validity of the lidar and our method in indoor tracking.

Key words: track-before-detect multiple extended target tracking clutter map track association