城市精细三维模型是智慧城市的关键基础信息，针对空地一体密集匹配点云数据精度不一、密度变化剧烈、同名地物特征相异导致的建筑物几何基元的高保真提取、拓扑关系的推理与恢复困难，本项目提出一种智能增广的空地一体密集匹配点云建筑物精细重建方法，主要研究内容包括：1）密度感知的建筑物平面基元提取，从密度变化的点云中提取完整平面基元；2）建筑物平面基元边界的多层次规则化，还原基元高保真边界及其正则关系；3）智能增广的基元拓扑恢复与模型重建，生成几何与拓扑最优的三维模型；4）原型系统与典型应用实验。本项目旨在显著提高空地一体摄影测量重建建筑物结构化模型的精细程度与自动化水平，为大规模城市精细三维建筑模型的高效生产和智慧城市建设提供有力支撑。

Accurate 3D city models are critical information for smart cities, which play an important role in city management, safety guarantee, and public services. Automatic building reconstruction by aerial and terrestrial photogrammetry is one of the major methods for 3D city modelling. However, the integrated aerial and terrestrial point clouds various largely in precision and point density, as well as the coverage areas for the same objects. Therefore, in this project, an accurate building reconstruction method from integrated aerial and terrestrial dense matching point clouds by intelligent augmentation is proposed. Major research contents include: 1) density-aware building primitive extraction, which extract intact building primitives from point clouds with large density variation; 2) hierarchical building boundary regularization, which recover building boundaries with high fidelity and multiple regularity; 3) intelligent augmented topology recover and model generation, which reconstruct structured models with optimal geometry accuracy and without topology conflict; 4) prototype system development and experiments. The project aims at improving the detailed richness, geometry accuracy, automation level of 3D building reconstruction for aerial-terrestrial integrated photogrammetry, and supporting the efficient reconstruction of large scale 3D building models in city areas.