Detecting new building construction using small unmanned aerial system

This is none-machine-learning study, but the proposed method is relatively sample and can be rapidly conducted in practice.

The small Unmanned Aerial System (sUAS) is an emerging approach to monitor new buildings. sUAS acquires ultra-high-resolution imagery which provides visual evidence and reduces the necessity of in-situ investigation. It offers greater potential for building change detection when two epochs of images of the place of interest are captured. This study takes the entire urban area of Longfeng Town, Hubei Province, China as a test site, where two sets of 0.05 m resolution sUAS images were acquired on March 23, 2017 and June 6, 2017, respectively. In this short time interval, the heightened structures of the existing buildings consist of most changes.

This study proposes a sensitive building change detection method by integrating the visual and elevation information from sUAS images. Dense point clouds were generated using sUAS images without control points. Two Digital Surface Models (DSM) are generated based on point clouds to detect elevation changes between two epochs. With true-color images, the improved Triangle Greenness Index (TGI) is used to mask out the natural changes caused by seasonal vegetation growth.

Lastly, multiple criteria are utilized to identify changes in buildings including new buildings on the ground and new stories atop current buildings. The experimental result reveals that over 93.3% of building changes, including 3 new buildings and 25 stories and structures added to existing buildings are detected, which proves the validity of the proposed method for local land-use enforcement. The proposed method takes 5 minutes to extract changes from orthoimages and DSMs of 2 km2, while manual monitoring is more than 40 times slower.

Ning, Huan, Xiao Huang, Zhenlong Li, Cuizhen Wang, and Duowen Xiang. “Detecting new building construction in urban areas based on images of small unmanned aerial system.” Papers in Applied Geography 6, no. 1 (2020): 56-71. https://doi.org/10.1080/23754931.2019.1707108