A Spatial Clustering Algorithm Based on Spatial Topological Relations for GML Data

Abstract

GML is an application of XML in geographic information system, used to store spatial data. In this paper, algorithm SCTR-GML is proposed for spatial clustering in GML data. Compared with other spatial clustering algorithms, SCTR-GML clusters spatial objects based on the spatial topological relations, while the reported algorithms like DBSCAN just cluster the spatial objects that are near to each other into one cluster. The algorithm SCTR-GML firstly creates indexes for the spatial objects described by GML, computes all the topological relations including contain, intersect, adjacent, and also proposes a novel method to measure the similarity between spatial objects which needs the interaction of users. The objects in one cluster may not be near to each other, but they have similarity in the spatial topological relations. Encouraging simulation results are observed and reported. The experiment shows that SCTR-GML is effective and efficient.