顾及分布特征和道路网约束的居民地综合方法研究

Collaborative map generalization, spatial pattern recognition and maintenance, m-n features mapping and relationship mapping cartographic generalization are much more difficult in map generalization and multi-scale representation. Thence, with the perspective of spatial data enrichment and collaborative map generalization, this research aims to deeply excavate the key theories and technologies of settlement generalization considering spatial pattern and road networks as constraints. On large scale, lane-level road cluster is recognized and typified to create road mesh. Vector domain mathematical morphology approach to settlement generalization is proposed considering both road mesh and building density. For linear pattern and grid pattern in urban buildings, a new typification method is proposed. Through process identification, the typification operator is regarded as a progressive and iterative process consisting of three basic operators including elimination, exaggeration and displacement.The three basic operators are effectively organized to achieve the function of typification, which has great advantages in the guarantee of initial position, spatial characteristics and spatial patterns including linear pattern and grid pattern. On medium and small scale, a distance constrained Voronoi model is proposed, which can avoid the spatial conflict, avoid excessive temporary solidification to some extent, and optimize the ability of spatial distribution preserving. An adjacency length method is proposed to identify road roundabout and road corner, and a quantitative importance measurement is given. According to the principles of settlement selection, an improved multi-constrained generic selection strategy is proposed, which takes several constraints into account including multi-scale quantity consistency, semantic rules, spatial distribution, spatial conflict, adjacent roads, road roundabouts and corners.

协同地图综合、分布特征识别与保持、以m-n型多元素映射和关系映射为代表的结构地图综合等是制约地图综合和多尺度表达的瓶颈。为此，从空间数据增强和协同地图综合的角度入手，顾及分布特征和道路网约束，对居民地综合的理论和关键技术进行深入研究。在大比例尺上，通过道路线簇识别与简化，以道路网眼为单位，提出矢量域数学形态学，制订顾及道路网和密度约束的居民地综合方法；针对城市建筑群中直线模式和网格模式的分布特点，提出一种典型化新方法，将其分解为删除、夸大、移位等基本算子组合的迭代过程，力求最大程度保持原始分布模式的定位精度、空间特征和分布结构。在中小比例尺上，提出点群选取中空间冲突避免和空间分布保持的方法，制定道路路口、拐角的识别与重要性定量度量方法；并基于重力模型提出点群选取的道路协同方法；综合考虑多尺度约束、语义约束、空间分布、空间冲突、邻近关联道路、道路路口和拐角等因素，提出多约束的点群选取方法。

本课题针对多层次协同地图综合、分布特征识别与保持、以m:n型多元素映射和关系映射为代表的结构地图综合等制约地图综合和多尺度表达的瓶颈科学问题，从空间数据增强的角度入手，开展了如下研究。. 1）结合分布特征保持等地图综合需求，提出了建筑群、道路、点状居民地、自然面群等要素的多种分布模式认知特征及一系列识别算法，进一步发展了空间分布模式识别的理论与方法。. 2）针对大中比例尺居民地，兼顾道路的分割作用和居民地自身密度影响，设计了多要素协同的居民地综合方法。提出了矢量域数学形态学的建筑物综合方法、基于双向区域增长算法的道路线簇识别方法、道路线簇多层次典型化方法等。. 3）针对分布模式明显的区域，设计了顾及分布特征的、多算子协同的城市建筑群综合方法。以结构化地图综合中m:n型多元素映射和关系映射最具有代表性的典型化算子为例，提出了一系列多算子协同的地图综合策略，重点解决了结构特征保持的难题。. 4）针对中小比例尺上点群居民地，提出了顾及多尺度约束、空间分布、空间冲突、邻近关联要素等诸多因素的一系列改进算法，研制了多算法、多要素协同的多约束选取算法。. 研究成果为地图综合、空间数据增强、数据挖掘等科学问题提供了理论和方法支持，可为地图制图、空间数据库建设、数据集成与级联更新、多尺度表达等应用提供强有力的技术支撑。按照既定的研究计划，项目组完成了研究内容。课题组发表论文12篇，其中《Geocarto International》等SCI检索4篇，《测绘学报》等EI检索5篇；学术会议报告/论文9篇次；培养硕士研究生导师1名；培养博士生4名、本科生2名；1人获全军优秀博士学位论文；1人获2020年度国家科技进步二等奖（序2）。较好地完成了项目预期目标。