基于激光雷达构建室内三维地图关键技术研究

发布时间：2018-04-15 14:02

本文选题：激光雷达 + 点云数据　；参考：《电子科技大学》2017年硕士论文

【摘要】：由于我国城镇化速率的持续增加,室内的公共建筑也不断增多,其内部空间格调也变得愈加复杂。在应急疏散救灾、修建智能建筑、基于位置的服务、城市反恐作战等广泛领域都会对室内建筑的真实空间分布产生了很大的依赖。与传统测量方式相比,激光雷达技术是获取室内三维数据非常理想的途径,因此,采用激光雷达技术获取室内空间真实三维信息,以此作为室内三维地图图层的数据来源,具有一定的理论意义和现实价值。本文以某一楼层和楼层中某一具体房间为例,研究基于激光雷达构建室内三维地图的技术流程,其主要工作如下:(1)点云数据获取及预处理:本文使用Leica ScanStation C10三维激光扫描仪获取室内点云数据,并结合该仪器所获得的点云数据是散乱的特点,采用一系列方法完成了点云数据的组织与索引、去噪、相邻站点的数据拼接配准以及降采样等预处理工作。(2)采取随机采样一致性(Random Sample Consensus,RANSAC)算法实现室内点云数据平面特性检测以提取室内主体框架,而且通过根据欧式距离的层次聚类实现室内目标各物体的正确分割。(3)采用Delaunay三角剖分实现各目标物体的三维网格重建,并通过新增三角形法完成网格孔洞修补以及采用拉普拉斯平滑光顺算法实现网格模型的平滑光顺,从而完成各目标三维网格模型的优化。然后通过Arc GIS将模型文件生成shapefile文件,接着通过ArcSDE创建地理数据库,再与SQL Server 2008相连接起来,连接成功之后将前面所得到的shapefile文件全部导入到地理数据库中,最后通过Arc GIS Engine进行二次开发,实现地图可视化系统。主要实现地图三维可视化、旋转、平移、缩放、可通过小地图掌握全局、大地图目标查询使其高亮显示并居中、对感兴趣点的标记与存储管理以及呈现所查询目标的字段属性信息等基本功能。
[Abstract]:Due to the increasing of urbanization rate, the indoor public buildings are also increasing, and the style of interior space is becoming more and more complex.In many fields, such as emergency evacuation and disaster relief, construction of intelligent buildings, location-based services, urban anti-terror operations and other fields, there will be a great deal of dependence on the real spatial distribution of indoor buildings.Compared with the traditional measurement method, the laser radar technology is an ideal way to obtain the indoor 3D data. Therefore, the laser radar technology is used to obtain the real 3D information of indoor space, which is used as the data source of the indoor 3D map layer.It has certain theoretical significance and practical value.Taking a certain floor and a specific room in a floor as an example, this paper studies the technical process of building indoor 3D map based on lidar.The main work of this paper is as follows: (1) acquisition and preprocessing of point cloud data: in this paper, Leica ScanStation C10 laser scanner is used to acquire indoor point cloud data, and the point cloud data obtained by this instrument is scattered.A series of methods are used to complete the organization and indexing of point cloud data.The preprocessing work of data splicing registration and down-sampling of adjacent stations. (2) Random Sample ConsensusRANSAC) algorithm is adopted to detect the plane characteristics of indoor point cloud data in order to extract the main frame of indoor body.And according to the hierarchical clustering of Euclidean distance to realize the correct segmentation of each object in the room. The Delaunay triangulation is used to realize the 3D mesh reconstruction of each object.The mesh mesh is repaired by adding triangle method, and the mesh model is smoothed by Laplacian smoothing algorithm, thus the three dimensional mesh model of each target is optimized.Then the model file is generated by Arc GIS, then the geo-database is created by ArcSDE, and then connected with SQL Server 2008. After the connection is successful, all the shapefile files obtained above are imported into the geo-database.Finally, a map visualization system is developed by Arc GIS Engine.The main realization of 3D map visualization, rotation, translation, zoom, can be master the global situation through the small map, large map target query to highlight and center,The basic functions of marking and storing points of interest and rendering the field attribute information of the queried target.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN958.98

【参考文献】