The prestigious National Geographic magazine reported
about the use of
UAVs
(drones) to photography the daily life
and behavior of lions (Thurston, 2014). Animals are not dis-
turbed, because they did not consider these devices as threats.
Israel (2011) proposed a technology based on geo-refer-
enced thermal images for detecting fawns in areas where they
graze to avoid undesired damage. Monitoring was carried out
from a base station where the strips of video imagery were
received from an octo-copter.
But, not only wildlife on the savannah or the jungle are
monitored with
UAVs
; these platforms have been used for
remote monitoring marine mammals. Some reasons to use
UAVs
for this purpose were reported in Hodgson (2015).
Indeed, surveillance of large marine mammals was carried
out in Koski
et al.
(2009) with an
UAV
of 3.1 m wingspan, 1.2
m long, and 18 kg maximum gross weight, equipped with
a video camera. Surveys of sea lions for counting popula-
tions and other observations were carried out on the coast of
the Aleutian Islands (Walker, 2012). Because of the adverse
environmental conditions, several platforms (rotary and fixed
wing, including hand and catapult launching) were tested
and evaluated for this type of environment.
Martin
et al.
(2012) developed statistical models to esti-
mate the distribution of some hidden organisms that appear
when some factor affects the environment. They first mod-
eled the statistical distributions with tennis balls including a
known number of them with high probability to be occluded.
The underlying idea is to establish a relation between a gradi-
ent and the distribution. With this approach, they proposed to
know the relationship between the distribution of manatees,
based on georeferenced images, and the water temperature,
which is measured in-situ and related with the captured im-
ages. An electric powereded
UAV
used was a hand-launched,
2.7 m wingspan aircraft weighing 4.5 kg, equipped with an
off-the-shelf commercial digital camera.
Hodgson
et al.
(2013) surveyed dugongs in their marine
habitat at different flight heights at inaccessible areas avoid-
ing unnecessary human risks. This
UAV
is fuel powered, 3.11
m wingspan weighing 13.1 kg, with maximum
TOW
20.0 kg,
equipped with a commercial digital camera.
Bears, deer, and foxes were surveyed in the snow in
northwestern Miyagi, Japan (Oishi and Matsunaga, 2014).
Registered aerial images were captured from an unmanned
helicopter for automatic detection and identification based on
relevant moving points.
Belugas and other baleen whales species were monitored
with a turret-based stabilized, low-cost multispectral imaging
system onboard
UAVs
(Schoonmaker
et al.
, 2008).
Automatic bird detection techniques based on pattern rec-
ognition analysis from images captured with an
UAV
were ap-
plied as remote sensing approaches in Abd-Elrahman (2005),
where spectral similarities were exploited.
Sarda-Palomera
et al.
(2012) monitored temporal changes
in breeding population size in a black-headed gull colony
from
UAVs
that allows for observation in barren or bleak areas
without disturbance. The
UAV
was a fixed-wing with a 1 m
wingspan, weighing 2 kg, and equipped with an electric
brushless 250 W pusher propeller. Take-off was achieved us-
ing a catapult launcher.
Evaluation of an off-the-shelf
UAV
for surveying flocks of
Canada Geese and Snow Geese was carried out in Chabot
and Bird (2012), where the goal was comparing photographic
counts from repeated flybys of geese.
Automatic bird counting of a common gull colony was
used in Grenzdörffer (2013) with two
UAVs
, a quad-copter and
an octo-copter with payloads of 1200 g and 400 g, respective-
ly. Each was equipped with individual commercial, stabilized
cameras. A supervised classification approach, based on the
size of birds, was the method used.
Flora
Components and current capabilities of small
UAVs
have been
developed specifically for wildlife and ecological surveys that
is currently in field use for a variety of applications (Watts
et
al.
, 2010).
Monitoring vegetation and flora in rainforest areas is of
special interest to control the vegetation status. This research
was carried out in Koh and Wich (2012) with a fixed-wing
UAV
weighing 650 g approx., equipped with a commercial still
CMOS
camera. The study site and the national park are part of
a broader Leuser Ecosystem that contains the last few contigu-
ous lowland rainforests in Sumatra. Additionally, thanks to
the high camera resolution, large mammals are also surveyed
(elephants and orangutans).
Georeferenced images were acquired and processed for
monitoring vegetation strips classified as shrubs, woody, and
tree savannah (Lisein
et al.
, 2013). An autopilot fixed wing
with a 1 m wingspan and weight of 2 kg was the platform used.
Urban Environments and Infrastructure
In complex urban environments, some challenges for
UAVs
are surveillance, tracking, traffic control, car counting, illegal
construction, or building observation. These scenarios are es-
sentially complex including occluded areas where even
UAVs
equipped with a unique camera cannot reach or have difficult
access. Different efforts have been made to solve these kinds
of problems, including multi-
UAV
systems (Semsch, 2009).
Qin
et al.
(2013) reported on the difficulties found in urban
environments for mapping and modeling. Georeferencing and
processing were carried out with an octo-copter equipped
with a
RGB
camera assessing its validity against commercial
and non-commercial software packages. This octo-copter was
also used in Küng
et al.
(2011
b
) where automatic image pro-
cessing was proposed to generate building models for a final
manual editing and refinement. The procedure was based on
the generation of a 3
D
point cloud (dense matching) which
was projected on the z-axis to detect the building facades and
compute the primary directions in the buildings. Driveway
surveillance and control represents an urban application
where
UAVs
have been used (Cummings, 2013).
Visual-based tracking and control in urban areas is an
important issue addressed in Campoy
et al.
(2009). Often, the
GPS
signal is lost during navigation in such urban areas; it is
necessary to develop strategies to overcome this problem in
remote sensing applications. Three platforms were tested: two
gas powered helicopters and one electrically powered rotor-
motion
UAV
. All were equipped were with visual cameras.
Road information detection is another area of interest
where
UAVs
can play an important role for traffic monitoring
and control in Salvo
et al.
(2014), where a
VTOL
quad-rotor
with a payload of 300 g is the platform used. Street detection
based on video strips using color images is another applica-
tion proposed in Candamo
et al.
(2009).
Image-based change detection techniques are useful for
identifying new configurations of urban areas caused by legal
or illegal constructions;
UAVs
are suitable for early detection
and actuation (Walker, 2012).
Roca
et al.
(2013) proposed the use of an autopilot octo-
copter for the inspection of outdoor facades in buildings for
both structural and energy considerations. They argued that
these platforms are especially useful when facades are of dif-
ficult access making unfeasible the full 3
D
characterization.
They used an
UAV
equipped with a Microsoft Kinect sensor
for the generation of 3
D
point clouds.
Bulatov
et al.
(2011) proposed a four-step procedure (ori-
entation, dense reconstruction, urban terrain modeling, and
geo-referencing) for obtaining georeferenced 3
D
urban models
from video sequences, where
DTM
extraction, segregation of
buildings from vegetation and tree modeling were the most
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
April 2015
309
