一种基于地磁感应的车型分类算法*

地磁传感器是近年来出现的新型路面车辆检测器，而车型分类是车辆检测器的一项重要功能。根据地磁传感器的性能特点，提出了一种车型分类算法。该算法首先从原始数据中提取特征，进而对特征数据进行聚类以确定检测器的最佳分类能力，最后以“离线训练，在线分类”的策略，利用神经网络进行实时的车型分类，在不过度增加硬件负担的情况下获得了比较好的效果。该算法有助于提高地磁车辆检测器的性能，从而为交通管控系统提供更丰富的基础数据。

Vehicle classification algorithm based on anisotropic magneto resistive sensors

The anisotropic magneto resistive(AMR) sensor is a new type of road vehicle detectors,and vehicle classification is its necessary function.This paper raised a vehicle classification algorithm after analyzing the characteristics of AMR.This algorithm firstly extracted the features from the original data,then determined the classification ability by data clustering,and finally did real-time vehicle classification by a neural network using the so-called off-line training,on-line classification strategy.This al...