Theory And Method Of Streets Progressive Selection

Aiming at streets selection, this dissertation focuses on the research of theory of progressive map generalization, information theory applied to map generalization, knowledge acquisition for map generalization, constraint based automated map generalization and map generalization software. The main studies include:●The development of map generalization in the first 40 years is reviewed in the form of keywords. The directions of development of map generalization are showed based on the opinions of the field expert. The advantages and disadvantages of the existing methods of selection of streets are analyzed. The conceptual framework of progressive map generalization and similarity rule are proposed. The key concepts are:progressive, step-length, update, operator, representation, algorithm and evaluation. An operator model for street and a progressive representation model for street network are presented based on Abstract Cell Complex and seven-operator model.●The development of information theory applied to cartography is reviewed. Motivated by the decision tree learning algorithm ID3, a novel method based on the information change is proposed. The existing quantitative measures of map information are analyzed. The communicative model for streets progressive selection is proposed based on the idea of matching. The properties of this model are proved. Some principles of selection of streets are explained. The streets which are shorter and with lower connectivity are eliminated. When a small block needs to be merged to one of its neighbor blocks, consider the smallest block first.●The knowledge engineering of map generalization is summarized. Knowledge acquisition is carried out using C4.5 algorithm. Six parameters street hierarchy, street function, length, degree, betweenness, closeness were selected. They describe the semantic, topological and geometric attributes of streets. Three groups of experiment were conducted for different street network pattern and different scale transition range. The association of street elimination with different scale transition, different street network pattern, and street’s attributes is investigated.●The theory and method of constraint-based automated map generalization is summarized. A progressive selection of street method is proposed based on the combination of the constraint and data enrichment. Ordered Generalization Tree structure is presented based on the Dynamic Decision Tree. This method brings together the map generalization knowledge and the data structure, in such a way that the generalization process can be implemented. The streets progressive selection is implemented by block merging.●A map generalization software model which is common Platform plus Toolbox is discussed. Function of Toolbox is implemented by Web service. The concept and architecture of Web Services are introduced briefly. The concept and typology of map generalization services is described in detail. The thesis finally gives some service examples developed based on the.NET platform.The chapters in this thesis are independent from each other. More than 50 figures are arranged in the dissertation to explain principles of theory research, or illustrate the experiment results.