一种基于不变矩和SVM的车型识别器设计  

Vehicle Model Identification System Based on Invariant Moments and Support Vector Machine

作　　者：范伊红[1] 韩爱意[1] 

机构地区：[1]河南科技大学电子信息工程学院,河南洛阳471003

出　　处：《软件导刊》2007年第11期91-93,共3页Software Guide

摘　　要：介绍了一个基于不变矩和支持向量机技术的车型分类器。不变矩对于平移、旋转、尺度变化有良好的不变性,能很好的表示2维图像形状特征。支持向量机(SVM)建立在结构风险最小化原理基础上,对车型识别这样的非线性、高维数的小样本问题有非常好的分类效果和学习推广能力。在提取车长、车高等特征的基础上加入汽车图像不变矩特征,试验结果表明,该系统有很高的识别率。This paper presented a car model classifier based on invariant moments(IM)technology and support vector machine(SVM).Invariant moments are powerful methods to describe two-dimension images due to its independence of displacement,scale and orientation.SVM,based on structure risk minimization principle(SRM),has good ability of classification and generalization for the nonlinearity,multi-dimension,and small-sample problems such as car model identification.The presented classification system extracts the car image invariant moments as a feature except for the normal features such as car length,width and height.Experiment results show that the system has very high identification performance.

关 键 词：不变矩 支持向量机 汽车车型识别 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]