SURVEY OF BUILT ENVIRONMENT IN THE ERA OF UAV From Aerial Photogrammetry to Point Cloud Classification

In order to further discover the potentials of UAV (Unmanned Aerial Vehicle) for built environment research, this article involves in drone aerial survey and its post-processing, with a special focus on point cloud classification. By operating UAV flying over villages at foot of Mount Tai, capturing images of the villages as first-hand materials, and conducting research with the help of 3D model reconstruction software, deep learning implements, GIS environment, the findings of research response the questions of the relationship between flight altitude, working efficiency, and 3D reconstruction quality, and how to utilize the deep learning tools for certain building classification. The solution to the second problem, also the most noteworthy contribution of this article, is achieved by training a customized point cloud classification model. This model can be used to identify point clouds of specific types of buildings, which is an advancement compared to the basic Automated Classification in ArcGIS Pro. The quality of point cloud recognition is also better than the latter. Potential application of this research could be reflected in the statistical work for certain types of buildings. In other words, this study plays an intermediary role between UAV-aided image gathering to further spatial statistical research.