Research On Supporting Data Models And Methods For Maritime Feature Generalization

In map generalization field, the research meets the change from geometry oriented generalization to geographic-feature oriented generalization. The maritime feature which lies in coastal zone is one of important natural geographical features and the generalization of maritime feature is a typical case of geographic-feature oriented generalization. Coastline, island cluster and sounding are correspondingly points, lines and areas at geometry level also representing geographical characteristics of different kinds of geomorphic units. Aiming at maritime map generalization, taking coastline, island patch and sounding as generalization objects, the dissertation focuses on the research of data model and methods supporting geographic-feature oriented generalization from both theory and practice viewpoint. This study includes the following aspects.1) The study explores the concept of geographic-feature oriented generalization. The paper distinguishes geographic-feature oriented generalization from geometry oriented generalization from the viewpoint of generalization data, objective and methods. Theoretical foundation supporting geographic feature oriented generalization are systematically summarized from four different viewpoint—generalization theory, geography, spatial cognition and computational geometry. Some theories in different disciplines including coastal geomorphology, Gestalt cognition principles and Delaunay triangulation model are introduced in the application of maritime feature map generalization.2) Focusing on the lineal characteristics, the study explores the coastline generalization from the point of view of geographic-feature oriented. The novel strategy firstly analyzes the geography meaning the cartographic line conveys, and then combines the geographic knowledge and data model to execute the generalization process, finally examines the generalization result by geographic knowledge.This study focuses on two aspects of the geographic-feature oriented coastline generalization strategy:generalization oriented coastline classification schema and the algorithm realization of geographic-feature coastline generalization. On the basis of analysis of the requirements of geographic feature oriented generalization, taking geological origin and graphic characteristics into account, a new coastline classification schema is proposed. In this schema, coastlines are classified into four categories and nine sub-categories. The study summarizes the characteristics of each kind of coastline from three different aspects, i.e. sinuosity, geometry character and the relation of other features. After that, taking the kind characteristic of coastline as the main generalization constraint, we give the suggested generalization approach for each kind of coastline.The study takes ria coastline as an example to elaborate the implementation of geographic feature oriented algorithm. At the geography analysis level, based on the analysis of ria coast origin and geographic characteristics, we come to the conclusion that ria coastline results from land erosion and the graphic of coastlines bears dendritic structure. At the algorithm realization level, we build the hierarchical estuary tree model with the aid of constrained Delaunay triangulation to represent dendritic pattern of ria coastline. With the support of hierarchical estuary tree model, we compute the parameter of estuary length, area and so on, and furthermore we judge the importance of estuaries through the integration analysis. On the basis of previous analysis, estuaries of less importance are deleted to simply ria coastline.3) The study explores pattern-aware generalization method for island selection. On the basis of the analysis of the geographic meaning the linear island alignment conveys, taking the preservation of linear island alignment into account, a novel island selection approach is presented. At the theoretical level, the Gestalt cognition psychology plays an important role in the recognition of linear island alignment by several perceptual grouping principles, namely the law of proximity, compactness, good continuation and others etc. At the algorithm realization level, we divide the selection approach into three parts:linear island alignment detection, quality evaluation of alignments and selection implementation. At the stage of pattern recognition, the proximity graph and MST graph is constructed to represent spatial structure of island cluster with the aid of Delaunay triangulation. Then the edge of MST tree is pruned successively according to three different Gestalt conditions in order to extract linear island structure. At the stage of quality evaluation, the alignment hull is generated to represent spatial structure of island alignment and parameters like compactness, extensibility and linearity, are defined to assess different Gestalt aspect of linear island alignment. On the basis of quantitative analysis, mental experiment is conducted to get qualitative description of linear alignments. Finally, at the selection stage, the spatial distribution characteristics of island patch, such as distribution range, density and patterns, is analyzed with the aid of Voronoi-like structures for islands. Based on spatial distribution analysis, the outer islands and island alignments are selected to construct the skeleton of the island cluster. Finally, pattern-aware generalization island selection is realized by the means of skeleton island selection and background islands deletion.4) The study explores the sounding generalization problem under multiple constraints. From the aspect of problem analysis, after various generalization constraints on sounding selection is summarized, we point out that generalization of soundings is constrained by several conditions at different levels(macro, meso and micro level), namely the preservation of terrain features, the preservation of spatial distribution properties, navigation safety and others. From the aspect of algorithm realization, a formal analysis of generalization constraints is conducted with the aid of Delaunay triangulation and Voronoi dirgram. On the basis of constraint formal analysis, the relative importance of generalization constraints is sorted according to the purpose of the map and geographic characteristic of cartographic region. Finally, the generalization of sounding is realized by the selection of important sounding which reflects the terrain character, distribution properties or adjacency relation of the sounding cluster and the thinning of background sounding.5) The study develops a maritime feature generalization system from the practical aspect. The data model, algorithm and generalization strategy presented in former chapters have been realized in this system. The characteristics of generalization system including code realization of data model and generalization procedure control are elaborated.