基于CityGML的室内导航网络自动生成方法研究

发布时间：2018-11-24 08:08

【摘要】：作为室内导航基础数据,导航网络的生成对于获取建筑物内部环境及表达物体位置十分重要。但是,传统的导航网络生成方法研究多基于建筑物二维平面图展开,缺少对三维几何信息和语义的综合运用。作为一种富含语义信息的三维室内模型,CityGML LOD4模型的出现为室内导航网络构建提供了全新的解决方案。本文从室内导航的实际需求出发,利用CityGML LOD4丰富的室内几何和语义信息,提出一种基于CityGML LOD4的室内三维导航网络自动生成方法。针对此过程中出现的特殊形状导航空间内路径生成错误问题,采用计算几何三维剖分算法对凹多面体进行凸剖分,并对室内导航拓扑网络进行了优化重构。实验结果证明,该方法所提取的室内导航网络,能够解决当前室内导航网络在构建过程中存在的自动化程度较低、忽略语义及三维几何信息、导航网络数据难以共享等问题,为室内导航与位置服务的更好实现奠定了基础。本文主要研究内容包括:(1)室内导航数据标准介绍。简要介绍了CityGML和IndoorGML,结合二者对室内空间的各类信息表达进行了详细的描述,以说明其在自动生成室内导航网络过程中所起到的重要作用。(2)室内导航需求分析与CityGML LOD4数据解析。从室内导航用例场景和需求出发探讨了CityGML LOD4中可用于室内导航的关键类,并分别介绍了可导航区域、空间连接、垂直空间和不可导航区域的具体解析过程。(3)三维室内导航网络的自动化构建方法。基于室内空间的网络化表达原理,介绍了从CityGML LOD4自动构建三维导航网络路径的方法,并针对其中存在着凹多面体中心点落于其外部的问题,采用相关计算几何算法对凹多面体进行凸剖分,并综合考虑几何和语义信息重构导航网络,最后基于CGAL实现了凹多面体的凸剖分。(4)室内导航网络自动化构建实验与导航网络的共享及可视化表达。利用三维语义建模与转换工具建立CityGML LOD4三维室内语义模型,基于三维室内导航网络自动化构建关键技术,采用相应的实验手段实现了三维导航网络的自动化构建,并导出为便于共享交换的室内导航数据标准格式IndoorGML,最终利用开源技术实现了三维模型和导航网络的在线可视化。
[Abstract]:As the basic data of indoor navigation, the generation of navigation network is very important to obtain the internal environment of buildings and express the position of objects. However, the traditional methods of navigation network generation are mostly based on the two-dimensional plan of buildings, and lack of comprehensive application of 3D geometric information and semantics. As a three-dimensional indoor model rich in semantic information, the emergence of CityGML LOD4 model provides a new solution for the construction of indoor navigation network. Based on the practical requirements of indoor navigation and the abundant indoor geometry and semantic information of CityGML LOD4, this paper presents an automatic generation method of indoor 3D navigation network based on CityGML LOD4. In order to solve the problem of path generation error in special shape navigation space, the concave polyhedron is subdivided by computational geometry algorithm, and the topological network of indoor navigation is optimized and reconstructed. The experimental results show that the indoor navigation network extracted by this method can solve the problems such as low degree of automation, ignoring semantic and 3D geometric information, and difficult to share navigation network data. It lays a foundation for better realization of indoor navigation and location service. The main contents of this paper are as follows: (1) Indoor navigation data standard introduction. In this paper, CityGML and IndoorGML, are introduced in detail to describe all kinds of information expression in indoor space. In order to illustrate its important role in the automatic generation of indoor navigation network. (2) Indoor navigation requirement analysis and CityGML LOD4 data analysis. Based on the scene and requirement of indoor navigation, this paper discusses the key classes of indoor navigation in CityGML LOD4, and introduces the navigable area and spatial connection, respectively. The analytical process of vertical space and unnavigable area. (3) the automatic construction method of 3D indoor navigation network. Based on the networked representation principle of indoor space, this paper introduces the method of automatically constructing 3D navigation network path from CityGML LOD4, and aiming at the problem of concave polyhedron center falling on its exterior, The concave polyhedron is subdivided by correlation computational geometry algorithm, and the navigation network is reconstructed by taking geometric and semantic information into account. Finally, the convex partition of concave polyhedron is realized based on CGAL. (4) the experiment of building indoor navigation network automatically and the sharing and visualization of navigation network. The 3D indoor semantic model of CityGML LOD4 is established by using 3D semantic modeling and transformation tools. Based on the key technology of automatic construction of 3D indoor navigation network, the automatic construction of 3D navigation network is realized by corresponding experimental means. The standard format of indoor navigation data, IndoorGML, which is easy to share and exchange, is derived. Finally, the online visualization of 3D model and navigation network is realized by using open source technology.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P208

【参考文献】