摘要
提出了一种新颖的利用5个埋入式应变传感器获得车型特征参数的多传感器行驶车辆分类系统.该系统利用传感器的特殊位置分布与车辆各轴到达传感器时刻的相关性,对采集的信息进行像素级融合,突出信号的特征部分,提高特征提取的准确率和精度.最后利用D-S证据理论组合轴数、轴空间和轴重等特征证据对车型进行分类.实验结果表明该系统对行驶车辆识别率超过95%.
A novel multisensor system for moving vehicle classification using the vehicle feature parameters gained by five embedded concrete strain gages was put forward. First, the collected information was pixel-fused according to the pertinence between the special locations of the sensors and arrival time of the vehicles＇ axles. Thus the feature sections were reinforced, and the exactness and the accuracy of feature extraction were increased. The feature evidences of axle number, axle space and axle weight of the vehicles were then combined by the combination rule of D-S evidence theory, according to which, the vehicle types were classified finally. The result shows that over 95 % recognition rate of moving vehicle were gained using the proposed system.
出处
《天津大学学报》
EI
CAS
CSCD
北大核心
2008年第2期194-198,共5页
Journal of Tianjin University(Science and Technology)
基金
黑龙江省交通厅基金资助项目(HJZ-2004-12)
关键词
车辆分类
多传感器系统
融合算法
INS组合规则
vehicle classification
multisensor system
fusion algorithm
D-S combination rule
