Abstract:
In today’s world, many applications require geolocated building information with accurate heights. Building heights can be used for various purposes, including estimating the number of floors, inspecting buildings that violate approved plans, assessing rental income, determining the number of people living in a place, and evaluating energy consumption. Obtaining accurate and reliable building heights has been a challenge. This study aimed to demonstrate how building heights can be estimated accurately using drone imagery. The methodology was tested in Juja sub-county, Kiambu County, Kenya.
Drone image data was used to generate Digital Terrain Model (DTM), Digital Surface Model (DSM), and normalized Digital Surface Model (nDSM) products, which aided in estimating building heights and floor numbers. The heights of buildings from drone data and ground survey methods were compared, yielding a correlation of 0.99. Similarly, a comparison between the number of floors from drone data and field observations showed a correlation of 0.92. Validation was also performed for the 2D aspects by comparing
the quality of the digitized orthophoto with the vectorized and regularized buildings extracted from the orthophoto through unsupervised classification. The intersection matching was 82%, which falls within the acceptable range for accuracy assessment. These results proved that drone data can sufficiently provide accurate building heights, saving human resources, money and time. Thus, applications requiring regular monitoring of building heights, especially during construction stages to determine compliance with
building regulations may consider the 3D reconstruction of overlapping aerial drone images.
Keywords:Drone Image Data, Digital Terrain Model (DTM), Digital Surface Model(DSM), Normalized Digital Surface Model (nDSM), Building Heights and Floor Numbers
