项目名称: 多尺度道路数据监督学习的匹配与选取更新方法
项目编号: No.41301523
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 天文学、地球科学
项目作者: 周琪
作者单位: 中国地质大学(武汉)
项目金额: 25万元
中文摘要: 地图数据库自动更新一直是地图制图学研究的国际前沿。我国目前理想的更新模式为:首先通过地图匹配方法对同一地区不同比例尺的新旧数据叠加分析,获取更新信息;再运用地图综合方法对增量信息进行缩编,其中最关键的问题即地图数据的匹配与综合。本项目以道路网络为例,探讨地图更新中适用于道路数据自动匹配和自动选取的自动化、智能化和普适性的新方法。针对现有道路自动匹配时因属性参数组合不当而导致误匹配的难题,提出了基于监督学习的自适应获取道路匹配时所需的最优参数组合;针对现有道路自动选取方法受区域和比例尺限制而普适性不足的缺陷,提出了具有较强通用性的基于监督学习的道路数据自适应选取方法。本项目的方法也可拓展用于其它线状地物如河流、铁路、等高线、管网等地图数据的自动更新。
中文关键词: 地图更新;道路选取;参数确定;机器学习;路网模式
英文摘要: Automated map database updating is always a hot research topic in the field of Cartography. An ideal method for updating is to first obtain updated information at small-scale datasets with the updated large-scale dataset by map matching; and then to make the updated information represent appropriately at small-scale datasets by map generalization. Indeed, map matching and map generalization are two key steps in this process. This study focuses on updating road networks, and aims to propose automated, intelligent and adaptive approaches for road matching and road selection (one of the operators for road generalization). On the one hand, in order to decrease the errors on road matching, supervised learning methods are employed to adaptively acquire the most appropriate combination of various properties (or parameters) needed for road matching; On the other hand, in order to overcome the limitation that the existing approaches for road selection may not always produce an appropriate representation, supervised learning methods are also employed to adaptively determine the appropriate selection representation for road networks. The proposed approaches in this study may also be applied to update rivers, railways, contours, pip networks and other linear datasets automatically.
英文关键词: Map Updating;Road Selection;Parameter Determination;Machine Learning;Road Network Pattern