Stroke Function-based Approach To Extracting Backbone Of Urban Road Network

With the continuous development of GIS application, more and more complex problem is dealt with in GIS and different levels or different scales of geographical data are required to support a GIS application. Current approach to multi-scale spatial database suffer from several drawbacks, including long construction period, isolated store of different scales of spatial data in the system, and difficulty in using and updating the data. Ideally, a variety of smaller scale spatial data can be automatically derived from a large scale spatial database. That requires model generalization technology in the GIS environment.The model generalization of urban road network is studied in this thesis. The goal of this study is to extract structural backbone of urban road network, which provides the top level for the construction of road network hierarchy. The paper analyzes the current theory and methods of model generalization of road network. Then the principle to follow, the factors to consider, and the criteria for result assessment in road selection is summed up and a stroke function-based approach for backbone extraction in urban road network is proposed. Stroke is designated as the selection unit in this approach. The shortest path analysis is performed to compute the network centrality of a stroke. The centrality, connectivity and length of a stroke are taken into account to calculate the comprehensive index of the stroke function. The greater the comprehensive index of a stroke is, the more significant role it plays in the road network, and the higher priority is given during selection process. A graph theory-based method is used to maintain the connectivity of the selected backbone network.To test the validity of the proposed selection method, an experimental system of road network generalization is designed and implemented based on ArcGIS Engine. Firstly, a number of typical local road networks are selected for comparative experiments. Then large urban road networks are selected for comparative experiments. The experiment results show that the proposed method based on stroke function can maintain the overall structure and connectivity of the road network, has good stability and versatility, and has significant advantages over the widely used selection method based on stroke length.