地形圖線形地物縮編之研究

Abstract

本文主要在以內政部出版之數值地圖為基礎，探討路網與水系網縮編之處理方式，以解決國內現行作業方式存在之問題，提高製圖縮編之自動化與一致性。 道路與河流縮編均涉及線形簡化與平滑處理，本文提出曲線形狀法，該法與視覺法及Douglas Peucker法比較結果，說明簡化效果除與人工編繪線形接近外，亦具應用於小比例尺簡化之潛力。 道路縮編程序中，包括資料前級處理、路網簡化以及線形之簡化與平滑等三個主要步驟。在道路資料前級處理中，本文發展路段搜尋法以自動化辨識路網內容，建立路段屬性，以提高後續之路網簡化自動化處理。路網簡化除考慮以路段長度之選取處理，還發展路段連接偵測法，以偵測誤刪之短路段，維持路段連接完整性。並將道路等級及不同地區道路選取方式納入準則設計。而線形簡化與平滑則以曲線形狀法為主。 在河流縮編上，探討單線河與雙線河縮編時所採取之對策，與道路縮編類似，其程序亦包括資料前級處理、水系網簡化以及相關之簡化與平滑技術等三個步驟。本文提出以不規則三角網結構，於河流資料前級處理中，設計自動化辨識河段、萃取河流中心線及計算河寬之方法；並以此結構設計水系網簡化中雙線河合併處理法。 本文應用提出之方法，將道路與河流縮編成五萬分之一及十萬分之一比例尺之成效，並與人工編繪之結果進行比較分析。

The thesis concentrates on the map generalization of roads and rivers to reduce manual operation and improve the automation and consistent of map production. The fundamental operation of generalization is linear simplification and smoothing. A shape simplification method on topographic map was presented in this thesis. The method is superior to visual method and Douglas Peucker method according to the evaluation of road networks, administrative districts and contour lines. There are three steps were included in the processing of road generalization that are data preprocessing, generalization of road networks and linear simplification and smoothing. A route searching method was proposed to automatically identify the contexts of road networks and establish the attribute of routes in data preprocessing. The metrology of road networks generalization used not only adopts the rule of route length but also develops a route connected detection method to ensure the logical consistent. Road classes and roads selection in different district were also considered in the rules of generalization. The procedures of river generalization were similar to road generalization including data preprocessing, generalization of river networks and linear simplification and smoothing. Constraint delaunay triangulations was applied to automatically identify branches and center lines of rivers in data preprocessing. The data structure was also conducted to design the aggregation of rivers in generalization of river networks to improve automation. The proposed methods in this thesis generalized roads and rivers to scale 1:50,000 and 1:100,000. An evaluation was applied to compare the generalized results with manual compilation.