42
scenes, pixel misalignments between images might be expected for these small 3D
features (e.g., trees, buildings, etc.).
While recent imagery acquired from larger manned platforms has demonstrated
a high degree of geometric quality, with standard and robust geometric correction
routines developed, correction of data from the newer Unmanned Aerial Vehicles
(UAVs or drone) platforms needs greater consideration and care. Specifically,
because of the weight requirements of the UAV platforms, lighter and lower quality
IMU and GPS units are fitted and therefore accuracy is lowered. Additionally, as the
acquisition process involves a large number of images, with each covering small
areas, an image matching process is required to create a single mosaicked image.
The overlapping regions of these images can also be used to build a high-resolution
DEM for the area, which can be subsequently utilised for the orthorectification of
the image mosaic (Jhan et al. 2016). It is recommended that ground control points
(GCPs) are acquired for ground targets unless differential GPS (dGPS) system with
real-time kinematic GNSS (RTK) or post-processed kinematic (PPK) are used during the UAV acquisition. Where GCPs are used, these will subsequently need to be
identified within the UAV imagery, which can be a time-consuming process.
However, with the latest RTK and PPK enabled GPS systems, pixel 9 locational
accuracies are commonly within ±5 cm in the x and y axis’ and ±10 cm in the z axis
without the need for manual intervention.
Optical Data
Optical sensors measure the amount of light that is reflected from the ground surface. However, between the ground surface and the sensor, there is an atmosphere
that contributes to the measured reflectance. There are various pre-processing stages
that can be applied, but removing the atmospheric and bidirectional effects is key to
providing a comparable and full standardised product. However, bidirectional
effects are commonly not corrected for (Nagol et al. 2015), as it can be difficult to
fully define the bidirectional reflectance distribution function (BRDF). For highresolution data, knowledge of the ground surface orientation at comparable resolutions or better is commonly not available. Bidirectional reflectance is the change in
the amount of light reflected due to the geometry of the acquisition, which is attributable to differences in the solar angles (e.g., with season and time of day) and sensor geometry (i.e., view angle of the sensor). These angles are with respect to the
ground surface, which themselves are defined with respect to the pixel resolution of
the imagery acquired. Therefore, for very high-resolution (VHR datasets, such as
acquired from a UAV), the orientation of individual leaves might need to be known
to correct for bidirectional effects within the image.
When energy (in this case, light) interacts with a medium, reflection, transmission or absorbance occurs. For example, as light from the sun interacts with plant
leaves, a proportion of this is reflected and transmitted and the remaining is absorbed.
It is the reflected component that is measured by remote sensing instruments.
P. Bunting
scenes, pixel misalignments between images might be expected for these small 3D
features (e.g., trees, buildings, etc.).
While recent imagery acquired from larger manned platforms has demonstrated
a high degree of geometric quality, with standard and robust geometric correction
routines developed, correction of data from the newer Unmanned Aerial Vehicles
(UAVs or drone) platforms needs greater consideration and care. Specifically,
because of the weight requirements of the UAV platforms, lighter and lower quality
IMU and GPS units are fitted and therefore accuracy is lowered. Additionally, as the
acquisition process involves a large number of images, with each covering small
areas, an image matching process is required to create a single mosaicked image.
The overlapping regions of these images can also be used to build a high-resolution
DEM for the area, which can be subsequently utilised for the orthorectification of
the image mosaic (Jhan et al. 2016). It is recommended that ground control points
(GCPs) are acquired for ground targets unless differential GPS (dGPS) system with
real-time kinematic GNSS (RTK) or post-processed kinematic (PPK) are used during the UAV acquisition. Where GCPs are used, these will subsequently need to be
identified within the UAV imagery, which can be a time-consuming process.
However, with the latest RTK and PPK enabled GPS systems, pixel 9 locational
accuracies are commonly within ±5 cm in the x and y axis’ and ±10 cm in the z axis
without the need for manual intervention.
Optical Data
Optical sensors measure the amount of light that is reflected from the ground surface. However, between the ground surface and the sensor, there is an atmosphere
that contributes to the measured reflectance. There are various pre-processing stages
that can be applied, but removing the atmospheric and bidirectional effects is key to
providing a comparable and full standardised product. However, bidirectional
effects are commonly not corrected for (Nagol et al. 2015), as it can be difficult to
fully define the bidirectional reflectance distribution function (BRDF). For highresolution data, knowledge of the ground surface orientation at comparable resolutions or better is commonly not available. Bidirectional reflectance is the change in
the amount of light reflected due to the geometry of the acquisition, which is attributable to differences in the solar angles (e.g., with season and time of day) and sensor geometry (i.e., view angle of the sensor). These angles are with respect to the
ground surface, which themselves are defined with respect to the pixel resolution of
the imagery acquired. Therefore, for very high-resolution (VHR datasets, such as
acquired from a UAV), the orientation of individual leaves might need to be known
to correct for bidirectional effects within the image.
When energy (in this case, light) interacts with a medium, reflection, transmission or absorbance occurs. For example, as light from the sun interacts with plant
leaves, a proportion of this is reflected and transmitted and the remaining is absorbed.
It is the reflected component that is measured by remote sensing instruments.
P. Bunting