摘要
为了提高车辆图像的识别率,提出了利用支持向量机(SVM)理论进行轿车车型识别方法。SVM能够解决线性及非线性分类问题,以较少的支持向量确定分类面,对样本数量及维数不敏感。基于颜色直方图及惯性比确定的图像特征具有平移、旋转和尺度不变性,可以用来确定SVM的最优分类面,并由此识别车型。
In order to increase recognized rate, a method for recognition of car model based on SVM isintroduced. Linear and nonlinear classifications could be solved with SVM, and the hyperplane could be defined by a few support vectors. SVM was insensitive to dimensions and numbers of training set. The features based on color histogram and ratio of inertial moment could keep their scale-invariance, rotation-invariance and translation-invariance, and got optimal hyperplane and recognize car model.
出处
《计算机工程与设计》
CSCD
北大核心
2005年第9期2453-2454,共2页
Computer Engineering and Design
关键词
图像
识别
特征
SVM
车型
image
recognition
feature
SVM
car model
