设施POI分布热点分析的网络核密度估计方法

设施POI(point of interest)在城市地理空间中往往聚集分布,呈现热点特征。对该类POI分布热点的分析大多采用基于欧氏距离的空间密度估计,忽略了城市空间通达、连接是沿着街道路径的事实,从而很难准确、客观地反映城市功能的热点布局。本研究针对该缺陷,利用基于网络路径距离的核密度计算方法确定热点的区域密度,并提出了一种简单、高效的网络分析算法。该算法扩展二维栅格膨胀操作,以一维形态算子的连续扩展计算POI在网络单元上的密度值,通过评价试验表明,该算法比现有算法具有更好的性能和可扩展性。通过实际POI数据分析发现,考虑街道网络约束的热点范围可凸显设施功能沿交通网络布局的空间特征,为区域规划、导航以及地理信息查询等应用提供有价值的空间知识与信息服务。

Network Kernel Density Estimation for the Analysis of Facility POI Hotspots

The distribution pattern of urban facility POIs (points of interest) usually forms clusters (i.e. "hotspots") in urban geographic space. To detect such type of hotspot, the methods mostly employ spatial density estimation based on Euclidean distance, ignoring the fact that the service function and interrelation of urban feasibilities is carried out on the network path distance, neither than conventional Euclidean distance. By using these methods, it is difficult to exactly and objectively delimitate the shape and the size of hotspot. Therefore, this research adopts the kernel density estimation based on the network distance to compute the density of hotspot and proposes a simple and efficient algorithm. The algorithm extends the 2D dilation operator to the 1D morphological operator, thus computing the density of network unit. Through evaluation experiment, it is suggested that the algorithm is more efficient and scalable than the existing algorithms. Based on the case study on real POI data, the range of hotspot can highlight the spatial characteristic of urban functions along traffic routes, in order to provide valuable spatial knowledge and information services for the applications of region planning, navigation and geographic information inquiring.