摘要
行人作为交通事故易受伤群体之一,其安全保障越发受到重视。结合车载激光测距仪实时采集的车辆前方障碍物距离信息,提出基于K-means算法的行人检测方法。首先对激光测距仪接收的距离信息进行报文解析,形成激光云点图。其次,对激光云点图进行预处理,消除冗余数据。再应用K-means聚类算法对前方障碍物进行分类,最后建立行人宽度模型甄别行人目标。试验结果表明,基于K-means聚类算法能从激光云点图中快速提取行人目标,为汽车主动安全及交通安全研究提供基础。
Pedestrians are one of the most vulnerable groups in road accidents, much more attention have been paid to the quarantee of pedestrain safety. Combining with range information of obstacles in front of a vehicle real-time acquired from on-board laser range finder, a pedestrian detection algorithm based on K- means is presented. First, the range data from laser range finder is parsed and laser point cloud diagrams are constructed. Then, the laser point cloud diagrams are preprocessed to reduce the redundant data. And the K- means clustering algorithm is utilized to classify various obstacles in front of a vehicle. Finally, the pedestrian width model is established to identify the target. The experimental result shows that the clustering algorithm based on K-means can extract pedestrians from laser point cloud diagrams, which can lay a foundation for the research of automotive active safety and traffic safety.
出处
《公路交通科技》
CAS
CSCD
北大核心
2014年第7期143-147,共5页
Journal of Highway and Transportation Research and Development
基金
国家自然科学基金项目(51108192
51208500)
中国博士后科学基金项目(2012M521824
2013T60904)
华南理工大学中央高校基本科研业务费专项资金项目(2012ZZ0100
2014ZG0029)
华南理工大学"学生研究计划"SRP(4564)
