灰度直方图与LDA的路况分类  

Research on traffic information classification based on a grey-histogram and linear discriminant analysis

作　　者：史晓鹏[1] 何为[1] 韩力群[1] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048

出　　处：《智能系统学报》2011年第4期328-332,共5页CAAI Transactions on Intelligent Systems

摘　　要：道路路况实时分类研究,是路况信息诱导系统的基础.通过对大量路况图像进行研究,提出使用路况图像中道路区域统计灰度直方图作为表征路况信息的特征向量,采用LDA(linear discriminat analysis)算法对高维特征向量进行降维,采用改进的K-近邻分类器实现对道路路况实时分类,并给出实际分类结果.实验结果表明,采用上述方法进行路况分类,其结果与交通管理部门给出的结果一致率达91.7%,对路况实时分类具有较高的实用价值.Research on real-time traffic information classification is the basis of traffic guidance systems. In this pa- per, a high dimension feature vector based on a grey-histogram sampling of a road region image was proposed. The classification algorithms of linear discriminant analysis （LDA） and an improved K-Nearest Neighborhood （K-NN） were adopted to reduce the high dimension vector and classify real time traffic information. The experimental results show that the proposed traffic information classification method can supply the same traffic information as what comes from the Beijing Traffic Management Bureau; the rate of identical information is as high as 91.7%.

关 键 词：路况分类 灰度直方图 LDA K-近邻分类器 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]