摘要： 针对车辆分类需求，提出一种基于误差椭圆的车型识别算法。利用背景差法去除车辆图像的不相关背景，从而分离出目标车辆，并对其进行识别和轮廓提取，通过平移、旋转和缩放车辆的轮廓边界，获得一个不相关的二维方差阵，将其与已知模板方差阵进行比较，以实现车辆分类。实验结果表明，该算法能获得较好的分类结果，满足实时性要求。

关键词: 车型识别, 误差椭圆, 智能交通系统, 轮廓提取

Abstract: According to the requirements of vehicle classification, this paper presents a vehicle type recognition algorithm based on error ellipse. It uses background differential method to remove the uncorrelated background of vehicle image, so to get the target image. Vehicle recognition and contour extraction is done by image analysis. A 2D coordinate uncorrelated variance matrix is got by translating, rotating and zoom operating the vehicle outlines. The uncorrelated variance matrix is used to compare with the vehicle classification template in advance and a vehicle is classified to the right type. Experimental results show that the algorithm can obtain good classification results, and meet the requirement of real-time.

Key words: vehicle type recognition, error ellipse, Intelligent Transportation System(ITS), contour extraction

中图分类号: 

• P207