中文摘要：

针对当前从建筑物设计图中提取通行结点网络的方式存在人工干预过多和效率较低等问题,该文提出了一种自动化构建室内通行网络的方案。在总结当前以CityGML LOD4格式存储的建筑物通行数据存在的各类典型问题之后,进行对应性的数据错误修正和补全;然后结合室内可通行空间特点,引入正交多边形分割算法和邻近关系分析算法,完成室内通行网络提取。实验结果证明,该方法所提取的基于CityGML LOD4文件的基本通行结点网络,能够在保持空间定性连通关系的基础上融合空间定量属性,满足人员室内路径规划应用的多重需求。

英文摘要：

In view of the problem that existing solutions suffer from large amount of manual operations and low efficiency, an automatic extraction solution for establishing basic pedestrian-oriented node graph from CityGML LOD4 files was proposed. This purpose was fulfilled by analyzing existing data quality problems and introducing the rectilinear polygon decomposition with spatial neighboring relationship analysis for CityGML LOD4 models. The experiment results showed that this solution could assure high effi- ciency and minimum human-interference at the same time.