基于K均值聚类与Bayesian分类器的车型识别方法  

Vehicle Classification Based on K Means Clustering and Bayesian Classifier

作　　者：蔡晓庆[1] 林培群[1] 徐建闽[1] 彭登[2] 

机构地区：[1]华南理工大学土木与交通学院,广州510640 [2]华南理工大学自动化科学与工程学院,广州510640

出　　处：《交通信息与安全》2011年第4期20-23,共4页Journal of Transport Information and Safety

基　　金：国家自然科学基金资助项目(批准号:50878088);广东高校优秀青年创新人才培养计划项目(批准号:N9100310);中央高校基本科研业务费专项资金项目(批准号:2011ZM0116)资助

摘　　要：不同车型经过感应线圈时,线圈检测器采集过来信号变化的波形不同,同类车型对应采集的波形中如波峰极大值、峰值位置、平均值、波峰数等特征值都具有相似性。根据感应线圈采集所得信号的特点,讨论了一种基于K均值聚类与Bayesian分类器的车型识别方法。学习训练中,通过K均值聚类算法把不同波形的特征值归类,再采用Bayesian分类器算法去获得分类器。实验结果表明,该方法具有较高的车型识别精度。When different types of vehicles pass over the inductive coil, the waveforms of the signal change from the inductive coil are different. For similar types of vehicles, the maximum values of wave crests, peak position, average value, wave crest number and other eigenvalues from the inductive coil are similar. According to the feature of the signal col- lected by the inductive coil, this paper discusses a method of vehicle classification based on the K means clustering and Bayesian classifier. During learning and training, the K means clustering algorithm is used to classify the different eigenvalues and then the Bayesian classifier algorithm to get the classification. Experimental results show that the method is highly accurate.

关 键 词：感应线圈 K均值聚类 Bayesian分类器 车型识别 

分 类 号：U491[交通运输工程—交通运输规划与管理]