A CONSTRAINT-BASED PROGRESSIVE GENERALIZATION MODEL OF URBAN STREET NETWORK

J. Tian

Wuhan University,School of Resource and Environment Science,Wuhan,China

yutaka-2010@163.com

Generalization can be divided into two types, namely model generalization and cartographic generalization. These two types are closely related, often model generalization being a pre-process of cartographic generalization. This paper aims to develop the model generalization and emphasize the progressive idea. The bases of this research are the theory of constraint-based generalization and some generalization models which use graph and tree structure to solve the generalization of urban street network. This paper presents a new generalization model for selecting important and characteristic streets in an urban street network. This model is based on constraints and a novel data structure. The constraints can be used as specification, guidelines and evaluation standard. The data structure is a supporting tool and essential for the automation of generalization. We show that this model can generalize the street network step by step in the horizontal dimension and represent multi-scale results in the vertical dimension. The model has some advantages that it produces predictable result, minimizes the deviation and maintains the integrity of the objects.

At the beginning, constraints which are used in generalization of urban street network are described as controlling factors to generalization, mainly including functional, graphical and structural constraints. Functional constraints identify the preserved streets during the process; Graphical constraints specify the minimum length of streets and the minimum area of blocks surrounded by streets; Structural constraints maintain the quantity of streets, the streets density and the size relationship among blocks and streets according to the map scale. Then characterization of the constraints is illustrated and some parameters are defined, including street hierarchy, street function, street connectivity, block area, street length, street quantity and street density. Further, a data structure is introduced in an attempt to partly circumvent the problem of urban street network generalization. It relies on a topological structure based on faces and edges. The face represents the block and the edge represents medial axis of the street. Based on this representation, the constraints are applied in the data structure to guide how to merge the blocks and maintain the structure of the street network. The process is driven by calculating the importance of blocks and streets surrounding them. This data structure turns out to be a set of trees. The presented data structure is suitable for progressive automated generalization. The proposed approach is validated by a case study.

Innovations of this research are: in theoretic level, it emphasizes the important role of the progressive idea in generalization. In methodological level, it introduces constraints to a data structure to solve the specified problem while considering the factors comprehensively.