摘要
针对室内场景下无源人体目标无法准确定位的问题,提出了一种基于激光雷达与单目相机融合的人体多目标室内定位系统。使用多个低成本固态激光雷达在不同位置采集点云,拼接出整个场景的点云;结合视觉传感器捕获人脸目标,通过预先标定的内外参数将点云和图像实现映射,计算出人体的空间坐标。在室内场景(5.4 m×1.5 m)和室外场景(5.0 m×5.0 m)下进行验证实验,结果表明基于激光雷达与单目相机融合的室内定位系统能实时定位多个人体目标,平均定位误差约1.0 cm。
To solve the problem that the passive human body target can not be accurately positioned for the indoor scene,a multi-sensor-fused indoor positioning system is proposed.Firstly,using multiple low-cost solid-state lidars,point clouds at different locations are collected and stitched together point clouds of the entire scene.Then,the point cloud is mapped into the face image captured by visual sensor through the pre-calibrated internal and external parameters.Finally,the spatial coordinates of the human body are calculated.The verification experiment is performed under the indoor scenario(5.4 m×1.5 m)and outdoor scenario(5.0 m×5.0 m),respectively.The results show that the indoor positioning system,which is based on the lidar and monocular camera fusion,can locate multiple human body objectives in real time,and the average positioning error is about 1.0 cm.
作者
熊亦威
李慧诚
周朝
杨爱英
冯立辉
乔耀军
XIONG Yiwei;LI Huicheng;ZHOU Chao;YANG Aiying;FENG Lihui;QIAO Yaojun(School of optoelectronics,Beijing University of technology,Beijing 100081,China;School of information and communication engineering,Beijing University of Posts and telecommunications,Beijing 100876,China)
出处
《自动化与仪器仪表》
2022年第5期11-15,共5页
Automation & Instrumentation
基金
国家自然科学基金资助项目(61971046,62075012)。
关键词
激光雷达
单目相机
室内定位
lidar
monocular camera
indoor positioning
