面向三维城市建模的多点云数据融合方法综述

多细节层次的三维城市模型是数字城市和智慧社会的关键空间数据基础设施，从基于稀疏点线特征的交互式半自动化建模到基于密集点云的自动化智能化建模已经成为国际学术界和工业界的热点前沿。由于立体城市空间结构的复杂性，多类型、多站点和多时相的点云数据融合处理是三维城市建模的基本途径，其基本思想是将具有不同视角、密度、精度、尺度、细节、时间历元等特征的多点云数据进行一致性融合表达与集成处理，建立可直接面向计算分析的智能化表达的多点云模型。归纳总结了多点云数据的主要特点，针对时空基准与精度、尺度、语义3个层面的一致性处理，分析了多点云数据融合的主要发展趋势，并凝练了面向三维城市建模的多点云数据融合关键问题。

Multiple Point Clouds Data Fusion Method for 3D City Modeling

Three-dimensional city models in multiple levels of details consists of the fundamental geospatial data infrastructure for digital city and wisdom society, feature based interactive modeling using sparse point or line features and automatic modeling based on dense point clouds have triggered interests in both academic and industrial communities. Because of the complexity of spatial structure of three-dimensional cities, fusion of multi-source, multi-view and multi-temporal point clouds is a critical issue of three-dimensional city modeling. The basic idea is to integrate multiple point clouds data, which have different characteristics such as angle of view, density, accuracy, scale, level of detail, time stamps, into the same coherent representation. This paper first summarizes the main characterstics of ubiquitous point clouds data, and then analyzes the major trend of multiple point clouds data fusion methods from three aspects, time-space datum and precision, scale, and semantics. Finally, critical issues of multiple point clouds data fusion for 3D city modeling are given.