基于DDAG支持向量机的汽车车型识别  

Recognitio of vehicle types based on DDAGSVM

作　　者：曹洁[1] 白艳红[1] 

机构地区：[1]兰州理工大学电信工程与信息工程学院,兰州730050

出　　处：《电子测量技术》2009年第6期88-91,共4页Electronic Measurement Technology

基　　金：甘肃省自然基金项目(0710RJZA060)资助项目

摘　　要：文章应用DDAG支持向量机方法对汽车图像进行识别。首先采用背景差分方法提取运动汽车,然后用Otsu法进行阈值分割、基于边缘阴影检测方法消除阴影,最后提取运动汽车的RST不变特征,输入DDAG支持向量机进行训练和识别。实验表明,基于DDAG支持向量机的车型图像识别算法具有很好的性能。A method for recognizing vehicle image based on the Decision Directed Acyclic Graph Support Vector Machine. First, background subtraction theory was adopted to extract moving vehicle. And then threshold is segmentate by Otsu method and moving shadow detection based on edge information. At last extracting the feature vectors of the object, then training a SVM to classify vehicle. An experiment showed that the algorithm based on DDAGSVM has better performance.

关 键 词：DDAG支持向量机 车型识别 RST不变特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]