IEEE Geoscience and Remote Sensing Magazine - September 2022 - 54

DATA ACQUISITION
UAV HS sensors have data acquisition concerns similar
to MS camera systems. In particular, they require careful
calibration and georeferencing to acquire object spectral
responses. HS cameras are typically calibrated and characterized
in a laboratory to determine lens fall-off corrections,
temperature effects, absolute radiometric calibration
coefficients, and the spectral response [180], [204]. The essential
steps in the calibration process include corrections
based on sensor calibration, compensation of illuminationrelated
effects, atmospheric correction, object anisotropy
correction, so-called bidirectional reflectance distribution
(BRDF) correction, and topography correction [180].
Solutions for atmospheric correction are mostly empirical
line-based calibration methods. In these, two or more panels
with a known reflectance are installed in the area of interest.
A linear transformation is estimated between the image gray
values and object reflectance of the panels. Direct calibration
technologies utilize onboard irradiance observations
to transform image radiances to reflectance [205], [206].
The advantage is that reflectance targets are not needed on
the ground, which is particularly useful in forested environments,
where suitable places for installing panels might not
exist. Furthermore, irradiance sensors provide valuable information
about changes in illumination when data sets are
captured under varying conditions [180], [201].
Alternative and supplementary solutions for radiometric
calibration are software based for atmospheric correction.
These can be based on estimating irradiance by using
atmospheric radiative transfer models [180], [195]. When
utilizing blocks of images, it is also possible to develop
block adjustment-based approaches for radiometric correction.
These techniques determine correction functions,
for example, utilizing radiometric tie points in overlapping
image areas, reflectance panels, and irradiance observations,
depending on availability. BRDF modeling can be
integrated into the process. These methods are thus capable
of producing radiometrically homogeneous sets of images,
optimally adjusting external irradiance data and image
block data to compensate for varying illumination conditions
[207].
The georeferencing solution depends on the sensor type.
2D spectral images can be georeferenced utilizing wellestablished
SfM techniques (see the " Image-Based Point
Clouds " and " Platforms " sections). These methods work for
sensors based on mosaic filters, e.g., Cubert cameras. In the
case of time sequential HS scanning, such as the Senop HSI,
precise georeferencing of the nonaligned spectral bands can
be carried out by using 3D band matching procedures [208]
or implementing a suitable dynamic sensor model in bundle
block adjustment [209], [210].
The most common HS imagers are push broom scanners
that are highly sensitive to flight dynamics. Their georeferencing
is based on either precise GNSS/IMU systems
[180], [197] or utilizing an integrated approach founded
on a lower-quality GNSS/IMU and 2D camera [196], [198].
54
Push broom scanners can also be integrated with lidar to
enable accurate georeferencing [199]. Point spectrometers
are the most challenging to georeference. The location and
size/shape of the spectral footprint of each measured spectra
are different and determined by the position, orientation,
FOV, integration time of the spectroradiometer, flying
height and speed of the UAV, and surface topography [203].
RADAR
UAV radar is being developed based on low-cost, small,
lightweight, and low-power sensors, and it is expected to
be used for close-range Earth observations with high revisit
frequencies and immediate responses. Research into
UAV radar is focused on system development [211]. Fitting
all the necessary equipment into lightweight, commercial
UAVs with relatively small payloads is a challenge since radar
systems are typically power and space demanding. Few
works have been targeted at tree and forest applications.
SENSORS AND PLATFORMS
Radar operates in the microwave wavelength to measure
ranges. It observes target areas during day and night, under
a wide range of weather conditions, and with wide swaths
and long ranges. It operates on different wavelengths. Short
wavelengths, e.g., the C and X bands, measure solid targets,
e.g., buildings, at high accuracy. Longer wavelengths, such
as the P and L bands, are useful when vegetation is present.
Low-frequency ultrawideband (UWB) millimeter radar has
good penetration capacity and offers potential for imaging
building interiors. Previous demonstrations of UAV SAR include
the P, L, and C bands [212] for civil and deformation
monitoring applications; X band [213] for calculating the
scattering characteristics of complex targets; P and C bands
for estimating the terrain height of a eucalyptus forest [214];
UWB [215] for snow coverage scanning; and W band [216]
to minimize the size and weight of the hardware, among
others, and different or full polarization combinations,
e.g., horizontal and vertical backscatters.
The onboard sensors typically include transmitting and
receiving antennas, a signal generation and recording unit, a
navigation system, and a control unit. Frequency-modulated
continuous-wave signals are common in current systems.
The antennas may be large in comparison to the UAVs, especially
for radar systems operating at longer wavelengths,
e.g., the L and P bands, which may negatively impact the aircrafts'
aerodynamic properties. Rotary-wing platforms suit
large antennas better because of their low platform speed.
The reported resolutions vary significantly. The resolution
was approximately, 1.5 m using the X band [217], 30 cm using
the C band [218], and 3 m using the P band [214] in the
slant range, and it was 0.5 and 5 m in single and multilook
images, respectively, using the P band [214] in the azimuth
direction. A planer resolution of better than 2 m obtained
using the C and P bands was reported [214].
Rotary-wing UAV platforms have great feasibility for agile
maneuvering. However, they have a much less stable flight
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE SEPTEMBER 2022
IEEE Geoscience and Remote Sensing Magazine - September 2022

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2022
