网络空间中线要素的核密度估计方法

A Kernel Density Estimation Method for Linear Features in Network Space

核密度估计(KDE)方法是分析点要素或线要素空间分布模式的一种重要方法,但目前线要素核密度方法只能分析线要素在二维均质平面空间的密度分布,不能正确分析交通拥堵、交叉口排队、出租车载客等线事件在一维非均质道路网络空间中的密度分布。本文提出了一种网络空间中线要素的核密度估计方法(网络线要素KDE方法),首先确定每个线要素在网络空间上的密度分布,然后根据网络空间距离和拓扑关系确定网络空间的线要素核密度与时空分布。以出租车GPS轨迹数据中提取的"上客"线事件为例,分析出租车"上客"线事件在网络空间中的密度分布,通过与现有方法比较的试验结果表明,本文提出的方法更能准确反映路网空间中线事件的分布特征。

Kernel density estimation（KDE） is an important method for analyzing spatial distributions of point features or linear features. So far the KDE methods for linear features analyze the features; spatial distributions by producing a smooth density surface over 2D homogeneous planar space, However, the planar KDE methods are not suited for analyzing the distribution characteristics of certain kinds of linear events, such as traffic jams, queue at intersections and taxi carrying passenger events, which usually occur in inhomogeneous 1D network space. This article presents a KDE method for linear features in network space, which first confirms the density distribution of each single linear feature, then computes the density distributions of all linear features in terms of distance and topology relationship in network space. This article extracts ＂pick-up＂ linear events from taxi GPS trajectory data and analyzes their distribution patterns in network space. By comparison with existing methods, experiment results show that the proposed method is able to represent the distribution patterns of linear events in network space more accurately.