对地观测技术的进步为城市三维建模提供了丰富的、不同粒度的数据源（如：影像数据、激光扫描数据），但不同平台获取的数据（我们称之为狭义的影像点云）的时空基准、表达方式和粒度精度方面均存在较大的不一致性，现有的理论与方法对其进行独立处理，无法进行优势互补和互联互通，难以构建可计算的三维城市模型，急需发展广义影像点云处理的理论与方法，从狭义影像点云走向广义影像点云，从可视化走向可计算。本项目重点建立广义影像点云的理论与方法体系。广义影像点云是指汇集摄影测量、激光扫描、众源采集等多平台空间数据，通过清洗、配准与整合，实现从多角度、视相关到全方位、视无关，建立具有基准一致、表达统一、多维感知和按需量测的广义影像点云，实现复杂场景的全覆盖描述；本项目发展广义影像点云的结构提取与语义标识方法，研究几何、关系、语义一体的多细节层次建模，建立从“广义影像点云构建—>几何结构识别与语义标识—>多细节层次建模”的理论与方法体系；以复杂街景环境为研究对象，以三维地籍、精准导航等应用为实例，实现从“逼真可视”到“精准感知”的跨越，推进GIS 由二维静态快速走向多维动态实时。

Earth observation technology enables to capture spatial data with different platforms rapidly. Nevertheless, the captured spatial data usually shows different resolutions, scales, and granularities, posing great challenges to process the data within a uniform framework. Hence, the complimentary between the captured spatial data can not be fully utilized for 3D urban model reconstructing and computing. The research grant aims to propose the theory and methodology of generalized imagery-point clouds (GIPC) to overcome the drawbacks of multiple platform spatial data robust harmonization and processing. The GIPC model has four attributes, namely, uniform coordinate reference, uniform representation, multi-dimensional sensibility, and measurement on demand. Three key problems will be investigated in the research grant, and listed as follows: 1. The construction of the generalized imagery-point clouds models, which aims to robustly generate the GIPC model with the spatial data from different platforms, 2. Sematic labeling and structures extraction from the generalized imagery-point clouds models, aiming to segment and classify the GIPC for objects detection, 3. Multiple levels of detail (LoD) representation of urban objects with correct topology, relations, and geometric details, aiming to build the computing 3D geometry models. Finally, 3D cadaster management and street-scene maps will be generated with the laser point clouds and imagery data from UAV platform and mobile mapping vehicles to validate the theory and methodology. The output from the research grant will build the pipeline of the generalized imagery-point clouds generation, the precise extraction of multiple features, and precise 3D modelling, and will provide services for many purposes and fields such as urban disaster management and smart city.