PE&RS April 2015 - page 285

platform during navigation. Some of these sensors are speed
or pressure gauges, inertial/angular measurement devices, and
also imaging sensors for SLAM . Thus, considering that such
sensors are specifically dedicated to navigation, they are no
longer considered part of the remote sensing system, although
they are critical for successful PaRS operations.
Additionally in remote sensing, different sensors work to-
gether as required for the application. Indeed, UAVs equipped
with accelerometers, magnetometers, gyroscopes (most times
embedded in an Inertial Measurement Unit, IMU), GPS, altim-
eters and cameras (optical, thermal, multispectral, or hyper-
spectral) associate each image with the corresponding GPS
location, altitude of the UAV, and orientation (pitch, roll, and
yaw angles), with the aim of obtaining geometric products,
i.e., 3 D mapping, geo-referenced images, and orthophotos.
Franceschini et al. (2010) designed a flexible architecture
for UAVs, aimed to enhance performance of large-scale metrol-
ogy instruments. Portability, flexibility, ease-to-use, and met-
rological performance are required of different technologies
onboard UAVs, including: optical and acoustic instruments,
mechanical/electromagnetic and inertial tracking systems.
Table 1 summarizes different sensors and instruments for
remote sensing onboard UAVs. Auxiliary sensors are identified as
the instruments onboard UAVs required to complete tasks or ap-
plications dedicated to specific sensors, which are also identified.
Sensors and instruments are in continuous improvement
and development, including the emergence of new technolo-
gies. This means that in the near future most of the existing
systems will change, becoming more effective. In this over-
view they are described as technologies with their capabilities.
Most of these technologies have been used in real-world
remote sensing applications as outlined later in the Applica-
tions Section. However, some other applications are waiting
for specific relevance, and are envisaged for potential future
applications.
Sensors and Technologies
Video Cameras in the Visible Spectrum
Video cameras are systems broadly used in UAVs for remote
sensing. This subsection deals with generic aspects related to
vision-based, onboard UAVs, operating in the spectral visible
range, i.e., from wavelengths from approximately 390 nm to
700 nm. Specific video systems are also considered in the Ap-
plications Section. Blyenburgh (2014) provided a review re-
garding imaging and range sensors, and Colomina and Molina
(2013) provided a representative list of small and medium
formats for visible band cameras.
Nawrat and Kuś (2013), inside the editorial of the Part II
related to the “construction of images acquisition devices used
in UAV applications,” addressed the problem associated with
image video acquisition related to the adverse flight condi-
tions derived from operations in day or night, adverse tem-
peratures, engines producing high frequencies and vibrations,
and unexpected UAV rotations because of wind variability
or gusts. It is suggested that the design of systems be robust
enough to deal with such situations, as well as proper systems
to be installed onboard the UAVs, i.e., with appropriate weight
and power consumption and with sufficient resistance against
adverse conditions. In addition, mechanical video stabiliza-
tion devices are to be considered ensuring that the camera
video system always points toward to the direction of interest
(Bereska et al., 2013). Stability analysis and geometric calibra-
tion of off-the-shelf digital cameras was addressed in Habib
and Morgan (2005). Figure 4 depict a visible camera, onboard
the helicopter HERO , as part of the sensor system installed on
a pan and tilt unit for stabilizing and targeting in the correct
direction. The objectives of this configuration are fire preven-
tion, detection, and monitoring (Martínez-de-Dios et al., 2007).
The need to cover large areas of vision led to the develop-
ment of systems equipped with capabilities such as omnidi-
rectional vision systems (Fraś et al. , 2013; Haus et al., 2013).
A commercial model with a CCD-based sensor is used with
an adjustable shutter speed ranging in 10 µs to 10 s, with ap-
proximately a 45º Field of View (FOV). The camera rotational
speed is 125 rpm capturing views of 360º.
Omnidirectional systems are suitable for surveillance ap-
plications. In this regard, Gurtner et al. (2009) investigated the
use of fish-eye lenses to increase the angle of view in aerial
photography, an application broadly used in remote sens-
ing. The lens distortion should be examined for its use with
low-quality cameras. The full suite of equipment has been
installed onboard mid- and small-sized (<10 kg) platforms.
The underlying concept is its use with small UAVs for remote
sensing tasks that cannot be achieved by satellites, such as for
monitoring of power lines or pipeline corridors.
Kim et al. (2010) designed an electro-optical system (EOS)
for small UAVs, able to track objects and recover 3D measure-
ments from these objects. The EOS consists of commercial
image acquisition systems integrated with the corresponding
servo-motors for pan and tilt orientation and stabilization. A
ground control station (GCS ) sends signals and receives infor-
mation to and from the EOS . The datasheet for EOS indicates a
weight of about 3.5 kg with a size of 178 × 178 × 269 mm. It is
installed onboard a platform with 2,050 mm wingspan, 79.5
dm2 wing area with an empty weight of 3.3 kg.
Li and Yang (2012) proposed the design of a UAV-based in-
telligent photography system able to capture image and video
stabilized sequences with any digital camera. The fixed-wing
UAV incorporated the following parameters, among others:
span = 3,360 mm; load = 4 kg; ceiling = 4.5 km. The images
are transmitted to a GCS . In the same way, Hodgson et al.
(2013) used commercial video-cameras, where the images are
transmitted in real-time to a GCS.
Undesired vibrations or rotations (pitch, roll, yaw) in UVA,
not detected by the IMU and affecting the image acquisition,
are compensated by developing software-based methods
for video stabilization, Fowers et al. (2007) and Wang et al.
(2012) used relevant features in the images (corners, lines)
for such purpose. This application was also addressed in
Buyukyazi et al. (2013) where the images are transmitted and
processed in a ground station. This was also the approach de-
veloped in Walha et al. (2013) based on point extraction and
matching techniques between consecutive frames.
Feifei et al. (2012) proposed a system with four cameras for
3D modeling based on triangulation from the overlapped images,
which are captured under different angles of view. The combi-
nation can reach 130°. Grenzdörffer et al. (2012) also proposed a
four-vision camera system, with weights of 80 g including lenses
with focal length of 9.65 mm. The radiometric and geometric
calibration problem (including inter calibration of the four cam-
eras) is addressed with the system onboard a quad-rotor.
Thermal Infrared Video Sensors
Differences between thermal and infrared sensors are due to
emitted and reflected energy, respectively. An infrared thermal
sensor detects radiant energy, based on the assumption that
objects with temperatures above absolute zero emit infrared ra-
diation as a function of wavelength and temperature. Accord-
ing to ISO 20473, the wavelengths of the spectral bands range
approximately as follows (Robles-Kelly and Huynh, 2013):
0.78 µm to 3 µm (near-infrared), 3 µm to 50 µm (middle-infra-
red), and 50 µm to 1000 µm (far-infrared). Some of these spec-
tral ranges can be integrated into multispectral or hyperspec-
tral sensors together with visible spectral ranges, considered
below. Colomina and Molina (2014) provided a representative
list of thermal cameras in UAVs. Infrared and thermal cameras
are devices capable of operating in adverse weather conditions
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING April 2015 285