【信息技术】【2008.09】自动车辆识别系统：使用汽车后视图和背光形状的方法

本文为葡萄牙里斯本大学（作者：David João Adão dos Santos）的硕士论文，共103页。

在现代社会，由于犯罪率高、交通事故多，人们的不安全感和威胁感日益增加。建立防御和预防机制的必要性鼓励研究开发车辆自动识别系统，例如，该系统应具有在一定距离识别车辆的能力。

本文报告的工作重点是两种不同方法的实施，两种方法都依赖于对车辆外部特征的分析，以确定车辆的制造商和型号。第一种方法利用了车辆的外部形状，特别是其尺寸和边缘；第二种方法考虑了根据车灯计算出的特征，特别是方向、偏心度、位置、与车辆牌照的角度及其形状轮廓。然后将这两种方法结合起来，以实现所提出的自动车辆识别系统。

实验结果表明，车辆外部特征可以作为一种可行的自动识别技术，在IST停车场获取的数据库测试上，其正确识别率为88%。

In modern society, due to the high crimerates and high number of traffic accidents the feeling of insecurity and threatis increasing. The need for the establishment of defence and preventionmechanisms has encouraged studies to develop an automatic recognition systems,which should for example, have the ability to recognize vehicles at a certaindistance. The work reported here focuses on the implementation of two differentmethods, both relying on the analysis of vehicle external features, to identifythe vehicle’s manufacturer and model. The first method evidences the externalshape of the vehicle, especially its dimensions and edges, while the secondconsiders features computed from the vehicle lights, notably the orientation,eccentricity, position, angle with vehicle license plate and its shape contour.Both methods are then combined in order to obtain the proposed automatic vehiclerecognition system. Experimental results indicate that vehicle externalfeatures can be exploited as a feasible technique for automatic vehiclerecognition showing a correct recognition rate of 88% on a database acquired atthe IST parking.

1 引言
2 当前技术水平综述
3 本文提出的车辆识别系统
4 软件实现与图形化用户界面
5 实验结果
6 结论
附录1 实验表格
附录2 IST数据库

下载英文原文地址：

http://page2.dfpan.com/fs/5lc3j2721629f163d31/

更多精彩文章请关注微信号：