Evaluation of unmanned aerial vehicles (UAV) high spatial resolution data to produce digital terrain model and visible spectral imagery

This article presents the results obtained with Unmanned Aerial Vehicle (UAV) high
spatial resolution data to produce a digital terrain model and visible spectral imagery. The
UAV developed in the Civil and Environmental Engineering Department at the University of
Brasilia (UnB) was used to obtain remote sensing visible spectral imagery with 6 cm spatial
resolution. The area of study, located in the city of Brasilia near Lake Paranoá, was imaged
with 500 images of 60% common area overlapping in the 1/1000 flight scale. Two Dense
Surface Models (DSM) and one orthomosaic of the study area were produced with Pix4D
software using a point cloud algorithm, the UAV`s GPS and altitude parameters and
geographic coordinates from ground control points (GPC) obtained using a topographic GPS
survey. The orthomosaic imagery obtained from the UAV was registered using GPC
measured in the area studied with topographic GPS using an ArcGis Georeferencing Module
and the following transformations were tested: 1st to 3rd order polynomial, adjust, projective
transformation and spline transformation. A total RMS deviation inferior to 10 cm was
obtained with spline transformation, which allows the UAV’s imagery to be used to update
the cartographic base with scales above 1/2000. In the ArcGis 10.2.1 system, spatial analyst
module algorithms (TIN generator, Raster Calculator, mathematical operators, etc.) were
used to transform the DSM into a digital terrain model (DTM) and compare it with Terracap’s
digital terrain model (DTM) cartographic base , showing that the UAV’s DTM presented high
geometric quality with an RMS error lower than one meter. The results obtained from the
UAV`s high spatial resolution imagery developed by the Civil and Environmental Engineering
Department at the University of Brasilia confirmed the high geometric and cartographic
quality of its data to be used in study areas of up to 10Km2, thus making it a quality option in
regards to the elevated cost of high spatial resolution satellite images