基于异向性磁阻传感器的车辆检测与车型分类

Vehicle detection and classification algorithm based on anisotropic magnetoresistive sensor

根据行驶车辆对地磁场的扰动效应,采用三轴异向性磁阻传感器HMC5883l设计了一种车辆检测与车型分类系统,通过软硬件相结合的方法克服了磁阻传感器输出信号基准值漂移对车辆检测的影响,针对车辆首尾相接与大型车辆的检测矛盾,提出了一种基于方差的多状态机自适应阈值车辆检测算法,并在检测出车辆的基础上,利用磁阻传感器三轴磁场信息提取车辆信号特征,选择算法复杂性低、分类能力强的BP神经网络对城市道路车辆进行分类,实际测试结果表明,本文车辆检测与分类算法测量精度高、鲁棒性强,可靠性及冗余度好。

A vehicle detection and classification system is developed with a three-axis anisotropic magnetoresisitive sensor HMC58831 according to the disturbance effect of moving vehicles on the geomagnetic field. The influence of the signal drift of magnetoresistive sensors on vehicle detection accuracy can be overcome by means of a software and hardware combined method. Aiming at the conflict between the detections of head and tail docking vehicles and large vehicles, a new variance-based multi-state machine adaptive threshold algorithm is proposed. On the basis that the vehicle has been detected, the three-axis magnetic field information of magnetoresistive sensors is used to extract the signal features of vehicles, and a BP neural network with less computation complexity and strong classification capability is used for classifying the vehicles on urban streets. The experimental results show that the proposed vehicle detection and classification algorithm features high measurement precision, good robustness and reliability, and reasonable redundancy.