基于形态分析的建筑群三维综合方法研究

项目名称： 基于形态分析的建筑群三维综合方法研究

项目编号： No.41201486

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 地理学

项目作者： 毛波

作者单位： 南京财经大学

项目金额： 25万元

中文摘要： 城市三维模型用途广泛，但因其数据量较大，使得基于网络的在线可视化成为制约其应用发展的瓶颈，因此需要研究综合算法，降低三维城市模型复杂度。已有的三维综合算法大多针对单个建筑物，难以有效处理复杂建筑群模型。为此，本项目将空间句法分析引入建筑群分布形态检测，并针对不同形态提出相应的综合算法。对于邻近建筑物实现基于屋顶的合并，对于离散分布的建筑物则采用典型化方法保留其分布特征。并基于视觉显著性原理，通过模式匹配算法，计算原模型与综合模型之间的视觉相似度，以评价综合算法结果。最后通过用户调查和眼动实验验证所提评价算法的准确度。本项目利用空间句法分析方法，研究三维建筑群分布特征，挖掘综合过程中的规则与约束，并将其运用到合并与典型化操作之中，丰富了三维地图综合的理论与方法；同时通过眼动实验，为综合结果评价打开了新的思路。

中文关键词： 三维地图综合；形态分析；三维建筑群；地图综合评价；多尺度建模

英文摘要： 3D city models can be applied in many applications, but visualization is a bottleneck because of the data volume. Therefore, generalization is required to decrease the complexity of 3D city models. Existing 3D generalization methods are mainly designed for signal building models, so they can hardly handle the building groups well. In this project, space syntax is introduced to analysis the morphological features of building groups for which different generalization operations are applied. For the nearby buildings, aggregation algorithm is proposed based on 3D roof structures merging, while the typification method is desgined to preserve the original morphological features of the distributed building group. Then based on visual salience, visual similarity between original and generalized models is calculated to evaluate and improve the generalization algorithms. Finally, the user survey and eye tracking experiment is carried out for verification of the proposed visual similarity evaluation algorithm. This project employes space syntax to study the morphological features of 3D building groups, and digs out the generalization rules and constraints for aggrigation and typificaiton. According to the project, 3D cartographic generalization theroy is enriched, and the generalization evaluation is improved by the eye tr

英文关键词： three dimensional map generalization；morphological analysis；three dimensional building groups；assessment of map generalization；multi-scale modeling