wingspan of 2000 mm with payload of 1.3 kg, was equipped
with a multispectral sensor with weight less than 1 kg.
Monitoring of wheat crops was evaluated and quantified
in Lelong
et al.
(2008) based on the computation of different
vegetation indices from images acquired in the visible and
near-infrared spectral bands. The work of Perry
et al.
(2012)
was the focus in this line for determining phenotyping traits.
Wal
et al.
(2013) used
UAV
for crop monitoring to overcome the
problem of using satellite in areas with a high density of clouds,
improving performance. An effective example is a lidar system
onboard a helicopter working together with a camera operating
in the visible spectrum for infrastructure inspections and crop
monitoring in unfamiliar scenarios (Merz and Chapman, 2011).
Sullivan
et al.
(2007) used a thermal infrared sensor, with
bands 7 µm to 14 µm, onboard a
UAV
to assess the water
stress in cotton canopy. Meron
et al.
(2013) analyzed different
technologies for crop stress detection based on measures of
temperature in foliage. The data were captured from thermog-
raphy sensors.
Córcoles
et al.
(2013) used a quad-rotor to determine the
leaf area index with a digital
RGB
camera onboard. Leaf area
index was studied in Duan
et al.
(2014) for three typical row
crops (maize, potatoes, and sunflowers). Data were acquired
in-situ
and from a
UAV
equipped with a 128-band hyperspec-
tral imaging sensor ranging from 350 nm to 1030 nm with a 5
nm bandwidth.
Verger
et al.
(2014) described an algorithm for determining
the green area index in wheat and rapeseed crops. A fixed-
wing
UAV
with 2 kg of weight, including the payload, was
used and equipped with a four
CMOS
-based system to acquire
images in four spectral bands based on interferential filters
operating at 550 nm (green), 660 nm (red), 735 nm (red edge),
and 790 nm (
NIR
) with well-defined sensitivities.
Saberioon
et al.
(2014) studied the status of nitrogen and
chlorophyll content in rice leaf by analyzing the visible bands
from images. In-situ ground-based results were checked against
the images captured with an integrated camera fixed-wing
UAV
.
Faiçal
et al.
(2014) designed a control strategy to apply
pesticides in crops involving
UAVs
and
WSN
deployed in the
ground.
Hung
et al.
(2014) apply a learning-based approach for
classifying three invasive weed species on the north-west
slopes of New South Wales, Australia. A filter bank was used
for feature extraction, and an explanation of the images cap-
tured was acquired with a high resolution commercial true
color camera onboard a hexa-copter with 1.5 kg of weight.
Burkart
et al.
(2015) used a hyperspectral flying goniometer
system, based on an octo-copter equipped with a spectrom-
eter mounted on an active gimbal for collecting multi-angular
hyperspectral data over wheat fields for vegetation indices
analysis based on
BRDF
.
Trees in Forestry
Important trees in forestry, where
UAV
-based applications
have been of interest with significant performances are citrus,
peach, olive, vineyards, and pistachio. Nevertheless, the ap-
plications reported in this overview can be easily extended to
other different kinds of trees. Biophysical parameters can be
estimated using different vegetation indices. Based on these
parameters, several image-based products can be obtained:
leaf area index, chlorophyll content, water stress detection,
health of plants, canopy analysis, photosynthesis, mapping of
areas (including 3
D
), or soil analysis among others.
Stress on citrus fruit due to water content was monitored
in Stagakis
et al.
(2012); they applied structural and physi-
ological indices for such purpose, obtained from a multi-
spectral camera operating in the visible and near-infrared
spectrum. Zarco-Tejada
et al.
(2009, 2012, and 2013e) and
Berni
et al.
(2009b) determined water stress, leaf biomass, and
chlorophyll content of the canopy in citrus, peach, and olive
orchards. They used a micro-hyperspectral imaging system,
2.7 kg weight, with six-bands and configurable filters with
different wavelengths centers (490 nm to 800 nm), which is
synchronized with an
IMU
for ortho-rectification. Also, a ther-
mal (FLIR) camera, weighing 1.7 kg with spectral responses
ranging in 7.3 µm to 1.3 µm, is used for measuring differences
of temperature between the ground and the crowns of trees.
The
UAV
is an auto-piloted helicopter with a fuel engine.
García-Ruiz
et al.
(2013) used a six-narrow-band multispec-
tral camera equipped with filter arrays at several wavelength
centers. Its weight is 700 g, and is installed onboard a six-copter
with a weight about 2,000 g. They compute vegetation indices,
which allow the analysis of loss of greenness in citrus trees.
A multi-spectral camera, acquiring high-resolution images
at 10 nm bandwidth in the visible and near-infrared, onboard
an
UAV
, was used in Guillen-Climent
et al.
(2012) to model
the fraction of active radiation in citrus and peach orchards
with unstructured rows, being useful in applications for pre-
cision agriculture.
Yield estimation in citrus (orange trees) was obtained with
a mini-helicopter and a machine vision system (MacArthur
et
al.
, 2006).
Berni
et al.
(2008) applied the factor known as “crown leaf
area index” in olive trees for chlorophyll content analysis
based on two camera-based instruments: (a) a multispectral
system (2.7 kg) with six individual sensors with interchange-
able optical filters; and (b) a thermal camera (1.7 kg) with
spectral response in the range 7.5 µm - 13 μm. These sensors
were installed onboard a rotary wing
UAV
with 7 kg of payload
and in a fixed wing
UAV
with payload of 5.5 kg.
Health canopy in olive orchards was studied in Zarco-
Tejada
et al.
(2013d) based on reflectance and fluorescence
analysis. Different vegetation indices are obtained through
hyperspectral cameras with spatial resolution of 30 cm and
260 spectral bands ranging in 400 nm to 900 nm.
Vineyards and grape vines have been crops of special inter-
est very early (Johnson
et al.
, 2003). Different sensors are used
for determining measures related to: chlorophyll function
and photosynthesis activity, leaf area indices, or plant health
status among others.
Berni
et al.
(2009a) used high-resolution thermal images to
obtain the tree canopy conductance and the crop water stress in-
dex in olive orchards. A hyperspectral scanner with 80 spectral
bands in the 0.43 µm to 12.5 μm spectral range was used on-
board an airborne system. Also an
UAV
was developed to carry a
thermal device in the infrared (
FLIR
) and a multispectral imaging
sensor (ranging in 7.5 µm to 13 μm). Calderón
et al.
(2013)
used multispectral (six-bands), thermal and hyper-spectral (260
bands) imagery for computing some indices (xanthophyll, chlo-
rophyll, carotenoids and blue/green/red) for determining the
water stress in olive trees caused by soil-borne fungus in some
regions. The
UAV
used for the multispectral and thermal acquisi-
tion had a 2 m wingspan for a fixed-wing platform at 5.8 kg
TOW
.
Hyperspectral images were acquired with a larger
UAV
with a 5
m wingspan for a fixed-wing platform having 13.5 kg
TOW
.
Relations between chlorophyll fluorescence and photo-
synthesis were analyzed by Zarco-Tejada
et al.
(2013
a
) in
vineyards. The results are validated against other terrestrial
systems, such as infrared gas analyzers sensors. An auto-pilot-
ed 2 m wingspan and fixed-wing platform at 5.8 kg
TOW
was
used to carry thermal and multispectral sensors and a wing-
span fixed-wing platform with up to three-hour endurance at
13.5 kg
TOW
. Zarco-Tejada
et al.
(2013c) used hyper-spectral
imagery to determine the carotenoid content in vineyards
related to the photosynthesis. A six-band multispectral cam-
era and a micro-hyperspectral imager with 260 bands are the
sensors used onboard the platforms.
The structure from motion was explored in Mathews and
Jensen (2013) to determine canopy in vineyard with a
RGB
296
April 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
