Revising self-best-fit strategy for stroke generating

Abstract

The strokes in road networks refer to a set of connected and non-branching road segments that follow the principle of good continuity. Generating strokes plays an important role in road network generalization, topological analysis, pattern recognition, and schematic map generation. In this study, the self-best-fit strategy for generating strokes was improved by prescribing road segment processing order based on the importance of the road segments. The importance of the road segments was determined by four parameters: length, degree, closeness and betweenness. The road networks of Detroit and Birmingham were used for experiments. Different stroke generating strategies were compared in terms of network functionality and visual recognition. In terms of network functionality, the improved self-best-fit strategy is superior to the every-best-fit strategy, and in terms of averages, it is superior to the self-best-fit strategy as well as the self-fit strategy. From a visual recognition perspective, the improved self-best-fit strategy tends to generate longer strokes with global property compared to the every-best-fit strategy.