摘要
提出了一种基于KL变换的汽车车型的识别方法.首先对车辆图像进行预处理,然后通过KL变换构造特征车,将样本投影到特征车子空间,得到的投影系数即为待识别车型的代数特征,最后利用K近邻法进行车型识别.实验结果表明了文中提出的车辆识别方法是简便、快捷、有效的.
This paper presents a kind of vehicle recognition method based on KL transform. Firstly, we preprocess the vehicle images. Then we build the eigen-vehicles by KL transform. The sample is projected onto the eigen-vehicle-subspace, and the coefficient of the projection is the algebra-feature of the vehicle. At last, we recognize the vehicle with KNN. Experimental results have shown good detective performance of the algorithm.
出处
《中南林学院学报》
CSCD
2004年第2期111-113,共3页
Journal of Central South Forestry University