基于统计分析和系统聚类的道路格网模式识别

Road Network Grid Pattern Recognition Based on Statistical Analysis and System Clustering

从格网模式道路网的基本特征出发,即格网模式的道路网中纵横交错的道路线相互近似正交,组成大小不一,且相互嵌套的格网单元。从城市地图上采集到100个区域的道路网作为研究对象,通过100个研究区域的大量实测数据,统计出格网模式道路网的所有道路线夹角值,并求出他们的平均值和标准差,利用非监督的聚类方法,识别出格网模式的道路网。本文方法效率高且易于运用在工程实践中,克服了以往的识别方法约束条件多、通用性受到限制的不足。

This paper starts from the basic characteristics of the grid road network,that is,the crisscross road lines in the grid road network are approximately orthogonal to each other,forming grid units of different sizes and nested with each other.The road network of 100 regions is collected from the city map as the research object.Through a large number of measured data of 100 research regions,the angle values of all road lines of the grid pattern road network are counted,and their average value and standard deviation are calcu-lated.The unsupervised clustering method is used to identify the grid pattern road network.The method in this paper is efficient and easy to apply in engineering practice,and overcomes the shortcomings of many constraints and limited generality of previous identifica-tion methods.