Abstract :
The concept of stroke originates from the Gestalt principle of good continuity. Generally speaking, strokes are non-branching, connected chains consisting of one or more linear segments. Strokes play an important role in topological analysis, map generalization, pattern recognition, and schematic map generation. The existing stroke building methods determine whether the concatenation of two segments at an intersection satisfy the concatenation criterion or whether the concatenation is useful for an entire stroke. However, these methods overlook the influence the concatenation may exert on the entire network. Following this idea, this paper first formalizes the stroke building problem and subsequently proposes a network functionality oriented method for stroke building based on global efficiency, which is often used as a measure of network functionality in complex network theory. This method considers the entire network from a network functionality perspective in each concatenation step. In each step, the method finds the concatenation that maximizes the global efficiency of the entire network from all possible concatenations that satisfy the adopted concatenation criterion. The road network of Shenzhen is used as study area for testing, comparison and analysis. The proposed method can usually generate unique stroke sets while guaranteeing the generation of the stroke sets with relatively high network functionality and a good visual perception.