摘要
利用激光雷达采集的距离信息,提出一种基于轨道横向与纵向几何特征的轨道检测方法。首先,在激光雷达的单次扫描数据中,根据轨道对周围环境的遮挡特性及高度跳变,完成对可能的轨顶关键点的检测;其次,构建轨道横截面模型,通过模型匹配的方式剔除前期的误检并确定最终轨道位置;最后,考虑轨道的纵向连续性和单轨之间的轨距、平行性特征,对轨顶检测点间进行关联形成轨道区段。研究结果表明:纵向单轨聚类及轨道横向关联可显著减少错误检测,最终轨道检测召回率为83.0%~97.5%,准确率为93.3%~97.9%。
A track detection method based on the lateral and longitudinal geometric features was proposed using the distance information recorded by a LiDAR sensor.Firstly,within a single scan,the possible rail head key points were detected on the basis of the background occlusion caused by rails and the height jump of rails.Then a rail profile model was built by model matching with which the false positives were eliminated and the positions of the rails were determined.Eventually,considering the longitudinal continuity of tracks and checking the gauge and parallelism between rails,the rail head key points were associated to form track sections.The results show that the longitudinal rail clustering and the lateral track association can decrease the false positives remarkably.The final recall of the track detection is 83.0%-97.5% while the precision is 93.3%-97.9%.
作者
郭子明
蔡伯根
姜维
GUO Ziming;CAI Baigen;JIANG Wei(School of Electronic and Information Engineering,Beijing Jiaotong University,Beijing 100044,China;State Key Laboratory of Rail Traffic Control and Safety,Beijing Jiaotong University,Beijing 100044,China;Beijing Engineering Research Center of EMC and GNSS Technology for Rail Transportation,Beijing Jiaotong University,Beijing 100044,China)
出处
《中南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第2期560-566,共7页
Journal of Central South University:Science and Technology
基金
北京市自然科学基金资助项目(4184096)
国家自然科学基金资助项目(61703034)~~
关键词
轨道检测
列车车载定位
模型匹配
关联
激光雷达
track detection
train-borne localization
model matching
association
light detection and ranging(LiDAR)
