摘要
提出了一种自动检测和识别汽车类型的方法。该方法分为两个阶段,首先,用Adaboost的学习算法检测图片中是否有正面的汽车并得到车辆的头部区域。第二,对车辆头部区域,提取SURF局部特征,并与数据库中的特征相匹配,跟据匹配的结果得到车辆的类型。在实验中,对821幅图片进行测试,其中包含48个不同类型的汽车,该算法正确识别率是81.6%。
In this paper;we present a method for automatic detection and recognition of car types.There are two main stages in this process:First,detect the car based on adaboost learning algorithm and get the region of car face.Second,recognition of the car based on SURF feature matching.In our experiment,the algorithm yielded a recognition rate of 81.6% when tested on about 800 images containing 48 different kinds of car type.
出处
《微型电脑应用》
2010年第8期51-52,56,共3页
Microcomputer Applications
