Road selection based on Voronoi diagrams and “strokes” in map generalization

Road selection is a prerequisite to effective road network generalization. This article introduces a novel algorithm for road network selection in map generalization, which take four types of information into consideration: statistical, metric, topological, and thematic at three spatial scales: macro-scale which describes the general pattern of networks, mezzo-scale that handles relationships among road segments, and micro-scale that focuses on individual roads’ properties. A set of measures is selected to quantify these different types of information at various spatial levels. An algorithm is then developed with the extraction of these measures based on Voronoi diagrams and a perceptual grouping method called “stroke”. The selection process consists of three consecutive steps: measuring network information based on Voronoi partitioning and stroke generation, selecting roads based on information extraction in the first step with strokes as selection unit, and assessing selection results. The algorithm is further tested with a real-world dataset: road network map at 1:10,000 scale and its generalized version at 1:50,000 scale in Wuhan, China. The result reveals that the algorithm can produce reasonable selection results and thus has the potential to be adopted in road selection in map generalization.