一种基于朴素贝叶斯分类的车道数量探测

Traffic Lane Numbers Detection Based on the Naive Bayesian Classification

针对浮动车数据采集成本低、采集速度快、覆盖范围广、蕴含丰富道路信息等特点,提出了一种基于浮动车数据的城市车道数量信息快速获取方法。该方法首先根据浮动车数据的空间分布特征,利用基于Delaunay三角网的密度聚类方法对数据进行优选;然后通过探测浮动车数据的覆盖宽度及其在道路横截面的分布状态,构建朴素贝叶斯分类器;最后采用朴素贝叶斯分类方法确定目标路段的车道数量。结果表明:该方法可以从低精度浮动车数据中快速获取车道数量信息,提取精度达到76.3%。

Aiming at characteristics of low cost,rapid collection speed,wide coverage and massive traffic information in collecting floating car data（FCD）,a rapid method to obtain urban lane number information based on FCD was proposed.Firstly,the density clustering method based on Delaunay triangulation network was used to choose the optimum data considering the spatial distribution characteristics of FCD.Then the naive Bayesian classifier was built through detecting the covered width of FCD and its distribution state on the transectbased road.Finally,the naive Bayesian classification was used to determine lane numbers in target road segments.The results show that this method can be used to obtain lane number information rapidly from FCD with low precision and the accuracy of lane number extraction is76.3%.