A Hierarchical approach for efficient DTM and building footprint extraction from UAV images
Open Access
- 30 September 2025
- journal article
- Published by Al-Qadisiyah Journal for Engineering Sciences (QJES) in Al-Qadisiyah Journal for Engineering Sciences
- Vol. 18 (3) , 265-271
Abstract
The utilisation of UAV imagery for the creation of digital maps is a compelling subject within the domains of photogrammetry and remote sensing. This work introduces a hierarchical method for automating the process of building, extracting, and outlining using images captured by drones. The flight plan should be initially planned to provide about 60-70\% overlap to guarantee thorough coverage and precise image matching. The altitude of the drone should be adjusted based on the intended resolution to achieve a balance between capturing fine details and covering a larger region. Next, the technique of photogrammetric image matching was utilised to generate orthophotos and the Digital Surface Model (DSM). Moreover, the Digital Terrain Model (DTM) was extracted from the DSM to differentiate non-ground objects, including buildings. Subsequently, building segments were identified by applying a threshold to the difference between the Digital Surface Model (DSM) and the Digital Terrain Model (DTM), enabling accurate extraction of building segments. Finally, building polygons were generated involving two stages: coarse and refined, considering the least squares adjustment process to guarantee accuracy and detail. The proposed method was applied to drone images captured on the campus of Al-Muthanna University in the southwest of Iraq. The qualitative and quantitative investigation indicated that the building polygons obtained were highly promising, with approximately one-meter geometric accuracy. Nevertheless, accurately differentiating between buildings and other human-made structures (such as tents) and resolving issues related to mismatching error still pose significant difficulties, highlighting the need for additional investigation and development.Keywords
This publication has 21 references indexed in Scilit:
- A Review of Building Extraction From Remote Sensing Imagery: Geometrical Structures and Semantic AttributesIEEE Transactions on Geoscience and Remote Sensing, 2024
- Generative AI design for building structuresAutomation in Construction, 2023
- Building footprint extraction from aerial photogrammetric point cloud data using its geometric featuresJournal of Building Engineering, 2023
- BuildMapper: A fully learnable framework for vectorized building contour extractionISPRS Journal of Photogrammetry and Remote Sensing, 2023
- Building outline delineation: From aerial images to polygons with an improved end-to-end learning frameworkISPRS Journal of Photogrammetry and Remote Sensing, 2021
- Automatic extraction of urban land information from unmanned aerial vehicle (UAV) dataEarth Science Informatics, 2020
- Building detection and regularisation using DSM and imagery informationThe Photogrammetric Record, 2019
- A deep learning approach to DTM extraction from imagery using rule-based training labelsISPRS Journal of Photogrammetry and Remote Sensing, 2018
- Building segmentation and outline extraction from UAV image-derived point clouds by a line growing algorithmInternational Journal of Digital Earth, 2017
- A New Building Extraction Postprocessing Framework for High-Spatial-Resolution Remote-Sensing ImageryIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016