Integrating general principles into mixed-integer programming to optimize schematic network maps

ABSTRACT

Schematic maps are popular for the representation of transport networks. Many automated methods have been developed to generate such maps. In these methods, optimization techniques work with various sets of constraints. Most of these constraints govern geometric properties of individual features. A few constraints address relationships among features, but none explicitly deal with the main structure of an entire network. We believe that preservation of main structure is the most important and preservation of relative relations is helpful. This is because human perception follows a global-to-local process. These constraints have recently been formed into four general principles, with two for global structure and two for relativity of features. This study develops an automated method by integrating these principles into the mixed-integer programming (MIP) framework. Experimental evaluations have been conducted with two sets of real-world transit networks. In comparison to the existing method, the proposed method has smaller fractal dimensions, better computational performance and higher scores in terms of clarity, recognition of major lines, visual simplicity and satisfaction. Therefore, it is concluded that the proposed method can generate schematic maps with improved clarity and aesthetics. The idea in this study is also helpful for the design of other visual representations.