Airborne Infrared Hyperspectral Imager for Intelligence, Surveillance and

Reconnaissance Applications
Eldon Puckrin1, Caroline S. Turcotte1, Marc-Andr Gagnon2, John Bastedo3, Vincent Farley2
and Martin Chamberland2
1
Defence R&D Canada Valcartier, 2459 Pie-XI Blvd N., Quebec, Qc, Canada G3J 1X5
2
Telops, Inc., 100-2600 St-Jean-Baptiste, Quebec, Qc, Canada G2E 6J5
3
PV Labs, 175 Longwood Road South, Suite 400A, Hamilton, On, Canada L8P 0A1
1
[email protected] / 2 [email protected] / 3 [email protected]

ABSTRACT
Persistent surveillance and collection of airborne intelligence, surveillance and reconnaissance information is
critical in todays warfare against terrorism. High resolution imagery in visible and infrared bands provides
valuable detection capabilities based on target shapes and temperatures. However, the spectral resolution
provided by a hyperspectral imager adds a spectral dimension to the measurements, leading to additional
tools for detection and identification of targets, based on their spectral signature. The Telops Hyper-Cam
sensor is an interferometer-based imaging system that enables the spatial and spectral analysis of targets
using a single sensor. It is based on the Fourier-transform technology yielding high spectral resolution and
enabling high accuracy radiometric calibration. It provides datacubes of up to 320256 pixels at spectral
resolutions as fine as 0.25 cm-1. The LWIR version covers the 8.0 to 11.8 m spectral range. The Hyper-Cam
has been recently used for the first time in two compact airborne platforms: a belly-mounted gyro-stabilized
platform and a gyro-stabilized gimbal ball. Both platforms are described in this paper, and successful results
of high-altitude detection and identification of targets, including industrial plumes, and chemical spills are
presented.

1.0 INTRODUCTION
Emerging applications in the civilian and defense context require sensors with state-of-the-art sensitivity and
capabilities. Among these sensors, the imaging spectrometer is an instrument yielding a large amount of rich
information about the measured scene. This powerful spectrometric tool is becoming more common in civil
applications such as search and rescue, geological surveys, pollution monitoring, forest fire detection and
monitoring and combustion studies. These sensors are also emerging as important assets for defense
operations through the role they can play for troop protection against chemical attacks [1], detection of mines
and unexploded ordnance (UXO) [2] and for military target detection [3].

Imaging spectrometers have unmatched capabilities to meet the requirements of these applications. The vast
majority of commercial-off-the-shelf airborne hyperspectral imaging sensors are of the dispersive push-broom
type. The Hyper-Cam sensor developed by Telops is unique in that it is an interferometer-based system that
operates in the 8-12 m longwave infrared (LWIR) spectral range. In principle, it can map regions of terrain
by using a step-and-stare technique if the pointing stability of the system is sufficiently reliable. The Hyper-
Cam is based on the Fourier-transform technology yielding high spectral resolution and enabling high-
accuracy radiometric calibration. The Hyper-Cam, a portable sensor, provides datacubes of up to 320256
pixels at 0.35 mrad spatial resolution and at spectral resolutions of up to 0.25 cm-1. The Hyper-Cam-LW has
been used in several ground-based field campaigns, including the demonstration of standoff chemical agent
detection [1].

RTO-MP-IST-112 19 - 1
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

The Hyper-Cam has been recently used for the first time in two compact airborne platforms: a belly-mounted
gyro-stabilized platform onboard a fixed-wing aircraft and a gyro-stabilized gimbal ball mounted at the front
of a helicopter. This paper describes both and presents successful results of high-altitude detection and
identification of targets including gases and chemical spills.

2.0 TELOPS HYPER-CAM-LW

The Hyper-Cam is a lightweight and compact imaging radiometric spectrometer. The spectral measurements
are performed using a Fourier-transform spectrometer (FTS), which incorporates a 320256 focal plane array
detector that can be windowed and formatted to fit the desired image size and to decrease the acquisition time.
Both, MCT and InSb focal plane arrays are used to cover the LWIR and the MWIR bands, respectively.
Spectral resolution is user selectable and ranges from 0.25 to 150 cm-1. This instrument generates a complete
spectrum of each pixel in the image, with each pixel having an instantaneous field-of-view of 0.35 mrad. This
field-portable sensor is shown in Figure 1. The instrument has two internal calibration blackbodies used to
perform an end-to-end radiometric calibration of the inf infrared
rared measurements. The sensor also has advanced
acquisition and processing electronics, offering the capability to convert the raw interferograms into spectra
using a real-time discrete-Fourier transform (DFT), and also offering the capability to apply the radiometric
calibration, generating for output real-time calibrated spectra in radiance units.

The control software has a user-friendly interface and provides real-time feedback to the operator. A
screenshot of the control software (named Reveal Pro) is presented in Figure 1. On the right, the non-
uniformity corrected broadband IR image is displayed with a greyscale code where black means low radiance
level and white high radiance level. The uncalibrated raw spectrum (or the interferogram as selected by the
operator) of a selected pixel in the image is displayed in real time on the left of the screen. The sensor has the
capability to change the focus of the IR image to produce a clear image from 3 meters up to infinity. The
Hyper-Cam was presented in detail in previous papers [1,4,5]. It has been successfully used during several
field trials.

Figure 1: Picture of the Hyper-Cam (left), a LWIR hyperspectral imager, and graphical user interface
(right).

19 - 2 RTO-MP-IST-112
 Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

3.0 BELLY-MOUNTED GYRO-STABILIZED AIRBORNE PLATFORM

The Hyper-Cam airborne system comprises the Hyper-Cam instrument along with several modules. The
primary function of the assembly is to compensate for the aircraft displacement and its angular pitch, roll and
yaw. It also adds accurate aircraft position and attitude data to the acquisition file metadata in order to later
geo-reference the acquired data. The following section describes the role of each subsystem to fulfill the flight
requirements.

In order to acquire the most useful infrared spectral ranges, the Telops airborne system allows mounting two
different Hyper-Cam instruments as shown in Figure 2. The Hyper-Cam-MW measures from 3 to 5.5 m
whereas the Hyper-Cam-LW measures the 8 to 11.5 m wavelength range. A single Hyper-Cam airborne
platform is also available. The optical bench includes a stabilization platform, the two Hyper-Cam
instruments, two Image Motion Compensation (IMC) mirrors, a GPS/INS unit and two visible boresight
cameras. All these modules are rigidly mounted on a high-stiffness base plate. This base plate is mounted on
the stabilization platform.

Figure 2: Illustration of the airborne configuration with two Hyper-Cams

Figure 3: Interferogram acquisition in airborne configuration

RTO-MP-IST-112 19 - 3
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

As illustrated in Figure 3, the IMC mirrors are used to compensate the airplane pitch, roll and forward motion
(independently for each Hyper-Cam sensor), while the stabilization platform is used to dampen the airplane
vibrations and to compensate the airplane yaw. The IMC mirrors are controlled by the navigation module
which receives and uses the information from two video trackers (one from each Hyper-Cam) and a GPS/INS
unit. The GPS/INS also enables ortho-rectification and geo-referencing of the collected data.

The Hyper-Cam instruments offer uncommon flexibility in adjusting their spatial, spectral and temporal
parameters. This flexibility proves to be invaluable for airborne applications where the flight parameters
impose severe restrictions on spectrometer operation.

4.0 GYRO-STABILIZED GIMBAL AIRBORNE PLATFORM

Camera systems in their aerial enclosures are given various names, including turrets, camera balls, and
gimbals. For the purpose of this discussion, the overall aerial sensor system, i.e. the Telops Hyper-Cam-LW
and the gyro-stabilized gimbal mount represented in Figure 4 will be referred as the gimbal. The primary
function of the mount carrying the camera is to provide a working environment for the sensor that contributes
the maximum amount to the task of the sensor providing a maximum of useful information quickly. Major
considerations for successful aerial imaging systems include environmental protection (shock/vibration,
temperature/moisture/pressure, wind loads), geo-spatial reference system, motion control (steering and
stabilization), payload capacity, and aircraft installation parameters. The PV Labs LDG-265 gimbal was
chosen for this application because it provided the optimum combination of performance and flexibility to
meet the system requirements for this application, as detailed below.

4.1 Aircraft installation parameters

For mission flexibility, it is desirable that the gimbal be able to work in a wide range of aircraft and
installations without significant re-packaging. To facilitate the widest possible range of aircraft installations,
the gimbal design ideally must provide a modular approach that permits a traditional internal mount looking
through an opening in the aircraft belly or be suitable for installation on the exterior structure via hard points,
pods, or pylons. For external mounting applications this necessitates an aerodynamic outer structure with
provisions for sealing, optical quality windows, and a range of motion suitable to provide the range of
coverage required by the mission CONOP.

4.2 Environmental protection

The gimbal structure is the host for the sensor system, and to permit success in capturing jitter free imagery
through a wide range of motion it must provide protection from wind loading, isolate shock and vibration, and
through design and component selection have sufficient sub-systems to monitor and control for effects of
temperature, pressure, and moisture. The range of possible aircraft requires a solution flexible enough to cover
the wind, safety, and manoeuvre loads of fixed wing aircraft as well as the relatively harsh vibration loads of
rotorcraft.

4.3 Geo-Spatial Reference System

Image stabilization systems require a reference system to compensate for angular motion and determine
absolute position information. Use of a common, high performance system for INS and steering/stabilization
permits real-time processing, tipping and cueing, and data indexing as well as minimization of cost and
system size/weight/power (SWaP).

19 - 4 RTO-MP-IST-112
 Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

4.4 Motion Control

Steering and stabilization are the key ingredients in achieving high quality imagery. Once environmental
protection has been provided, and precise geospatial information is available to the control system, the design
of the servo control system provides the final piece of the equation. 3 axes of stabilization is required for high
resolution, and sensor stability of less than 5 Radians is the target to achieve image motion blur less than
50% of pixel pitch in most cases. Motion control systems also need to provide solutions for Forward Motion
Compensation in both linear and orbital flight paths to eliminate blur and provide greater sensor integration
time. Tools to provide flight path control for the sensor through Geo-Steering also lead to more robust image
acquisition independent of aircraft flight path. For systems operating at or near nadir, it is imperative to
arrange control system axes to prevent gimbal-lock at nadir. The order of steering axes to permit this is
roll/pitch/azimuth, whereas most traditional ISR gimbal systems offer only 2 steering axes in Azimuth/Pitch
configuration.

4.5 Payload Capacity

To achieve maximum useful information on targets, it is desirable to provide a system with a multi-spectral
sensor solution bore-sighted to a common reference, and to maximize the number of sensors resident in a
single collection system. Sensors are continuously evolving, and must have the option of adopting a next
generation solution without significant cost and schedule implications. To provide this flexibility, the gimbal
system was designed on an easy-to-access open architecture which accommodate for relatively large payload
volume and modular sub-system components. As mission definition dictates or permits an increase or
reduction in overall system SWaP, the gimbal system must be able to follow without incurring significant
cost/schedule/performance risk.

Figure 4: The Telops Hyper-Cam in a PV Labs Look Down Gimbal

RTO-MP-IST-112 19 - 5
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

5.0 DETECTION AND IDENTIFICATION OF SUBSTANCES

To test the Hyper-Cam with the belly-mounted gyro-stabilized platform and with the gyro-stabilized gimbal
mount, two experiments were conducted. The experiment with the belly-mounted gyro-stabilized platform
was carried out on March 25, 2009 using a fixed-wing aircraft in which the airborne Hyper-Cam was able to
acquire measurements over a Canadian industrial site under clear sky conditions. The experiment with the
gyro-stabilized gimbal ball mounted on a helicopter was conducted on August 30, 2011 in summer conditions
at Defence R&D Canadas premises at Valcartier.

5.1 Detection algorithm

The Generalized Likelihood Ratio Test (GLRT) detection algorithm was used to identify the gas plume and
other chemical targets in these experiments. The GLRT algorithm has been used extensively by DRDC in
previous work, and a detailed description of the method is given in references [6, 7]. Essentially, the GLRT is
described mathematically by:

where the numerator represents the squared norm of the measurement, m, projected out of the background
space, B, and the denominator represents the squared norm of the measurement projected out of the
background + signature space, S. If the measurement does not contain the signature of interest, the result of
the GLRT is approximately one.

5.2 Industrial site with the belly-mounted gyro-stabilized platform

Measurements of an industrial site were obtained on March 25, 2009 during clear sky conditions. A colour
image of the site showing a prominent stack is presented in Figure 5A and the LWIR image measured by the
Hyper-Cam is shown in Figure 5B.

Chimney
(A) (B)

Figure 5: Images of the industrial site obtained from Google Earth (A) and with the airborne Hyper-
Cam-LW (B).

19 - 6 RTO-MP-IST-112
 Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

The LWIR image of the industrial site, shown in Figure 5B, measured with the airborne Hyper-Cam, was
analyzed for gaseous emissions using the GLRT algorithm. Figure 6 shows a summary of the detection and
identification results in the proximity of a large chimney. The measured radiance spectrum and SO2 gas
signature are represented in the upper panel of Figure 6A. It is evident that the SO2 is present even in the
unfiltered airborne Hyper-Cam measurement. The lower panel of Figure 6A also shows an excellent match
between the projected measurement and signature results. The output of the GLRT filter is shown in Figure
6B; the SO2 is detected well above the threshold value. The plume is clearly identified in the image of Figure
6C at the opening of the chimney. The corresponding ROC curve for the SO2 detection is given in Figure 6D,
where the detection probability is high compared to the probability of detection error [8].

(A) (B)

SO2
(C) (D)
PD

PE
Figure 6: The detection and identification of SO2 gas at the industrial site using the GLRT algorithm.
Measured radiance and SO2 signature (upper panel), and projected measurement and SO2 signature
results (lower panel) (A). Filter output showing the clear detection and identification of SO2 well
above the threshold value (B). Identification of the plume in the LWIR image (C). ROC curves for SO2
for a different numbers of basis vectors (D).

5.3 Experimental site with the gyro-stabilized gimbal

A target site was set up at Defence R&D Canada Valcartier to test the performance of the Hyper-Cam sensor
mounted in the gyro-stabilized gimbal, which was flown at altitudes of 300, 1000 and 3000 m on August 30,
2011 under clear-sky conditions.

RTO-MP-IST-112 19 - 7
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

The area containing the interrogated targets is shown in Figure 7A. Benign chemical targets consisting of
ammonium sulfate (AS) powder at different fill factors (12.5 %, 25 % and 100 %) were set up in 2m 2m
arrays. The area also included a controlled gas release apparatus intended for Freon (F-134a) dissemination.
The area contained other targets such as hotplates, acetone and methanol liquids and disturbed earth.

Figure 7: Target area as seen from flight altitude showing interrogated targets (A), 25% fill-factor
ammonium sulfate (AS) target (B), 100% fill-factor ammonium sulfate target (C), and gas release
apparatus on tripod (D).

Hyperspectral data measured for the scene at the different flight altitudes from the helicopter platform were
analyzed for the presence of ammonium sulfate powder using the GLRT algorithm. Figure 8 shows the
infrared broadband images of the interrogated area in black & white. The detection and identification results
from the GLRT algorithm are shown as a colored overlay pattern over the broadband image. The colored
pixels are those that showed positive detection of the interrogated material.

Figure 8: Broadband infrared image of the interrogated area with the colored overlay detection and
identification results of ammonium sulfate powder spills. The area is shown for altitudes of 300 m
(A), 1000 m (B) and 3000 m.

19 - 8 RTO-MP-IST-112
 Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

The experiment shows successful detection and identification of the ammonium sulfate powder from the
300 m altitude for all fill factors (grids of 12.5 %, 25 % and 100 % respectively). The imaging quality of the
Hyper-Cam-LW sensor allows for the clear distinction of the different fill-factor grid patterns and without
false alarms. The 1000 m and 3000 m altitude measurements provide successful detection and identification
for the 100 % fill-factor grid only. The ammonium sulfate spectral features for the lolower
wer fill factor of other
grids (12.5 % and 25 %) could not be differentiated from the surrounding area.

Figure 9 shows the results of hyperspectral data measured and exploited for the scene at an altitude of 300 m
using an acetone target consisting of a glass container filled with the liquid. The acetone was detected and
identified at a flight altitude of 300 m, as shown by the colored pixels (B) and the detection plane (C) of
Figure 9, but not at higher altitudes due to the small target size.

Figure 9: Picture of acetone target (A). Detection and identification of acetone from an open source
at 300 m altitude (B, C).

Figure 10 shows the results of hyperspectral data measu

measured
red and exploited for the scene at an altitude of 300 m
with a methanol target consisting of a glass container filled with the liquid. The methanol was detected and
identified at a flight altitude of 300 m, as shown by the colored pixels (A) and the detection plane (B) of
Figure 10, but not at higher altitudes due to the small target size.

Figure 10 Photo of methanol target (A). Detection and identification of methanol from an open
source at 300 m altitude (B,C).

RTO-MP-IST-112 19 - 9
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

The dissemination of F-134a gas could be detected and identified using the GLRT algorithm at all flight
altitudes, as shown in Figure 11. The Freon gas was released at a rate of 100 L/min during the measurement
period.

Figure 11: Broadband infrared image of the interrogated area with the coloured overlay detection
and identification results of a Freon F134-a release at an altitude of 300 m (A) and 3000 m (B).

6.0 CONCLUSION
The Hyper-Cam sensor was installed onboard a fixed-wing aircraft and a helicopter using a belly-mounted
gyro-stabilized platform and a gyro-stabilized gimbal ball, respectively. The fixed wing aircraft was flown
over an industrial site in which sulphur dioxide plumes were readily detected and identified using the Hyper-
Cam. The gimbal-mounted Hyper-Cam onboard the helicopter was flown over a site consisting of benign
chemical powders and gas plumes. In this case, the sensor was capable of detecting ammonium sulfate and
plumes of F-134a, along with methanol and acetone at lower altitudes.

The next steps will involve trying to co-add datacubes in order to further reduce the NESR of the sensor
system, which will depend on having sufficient stability in the control of the pointing mechanism. The
possibility to temporally co-add spectra is something not offered in typi
typical
cal pushbroom sensor systems.

7.0 REFERENCES
[1] Farley, V., Chamberland, M., Lagueux, P., et al., "Chemical agent detection and identification with a
hyperspectral imaging infrared sensor," Proceedings of SPIE Vol. 6661, 66610L (2007).

[2] J.E.McFee, C.Anger, S.Achal and T.Ivanco, Landmine detection using passive hyperspectral imaging,
Proc. SPIE Conference on Chemical and Biological Sensing VIII, Vol. 6554, Orlando, FL, USA, 9-13,
April, 2007.

[3] M. Shimoni, F. van der Meer, and M. Acheroy, Thermal imaging spectroscopy: Present technology and
future dual use applications, Proceedings 5th EARSeL Workshop on Imaging Spectroscopy. Bruges,

19 - 10 RTO-MP-IST-112
 Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

Belgium, April 23-25 2007.

[4] Puckrin, E., C.S. Turcotte, Lahaie, P., Dub, D., Farley, V., Lagueux P. and. Marcotte F. and
Chamberland, M., Airborne measurements in the infrared using imaging hyperspectral sensors,
Proceedings of SPIE Vol. 7324-46, Defense & Security 2009, Orlando, Florida, April 2009.

[5] Vallires, A., Villemaire, A., Chamberland, M., et al., "Algorithms for chemical detection, identification
and quantification for thermal hyperspectral imagers," Proceedings of SPIE Vol. 5995, 59950G (2005).

[6] Manolakis, M., Marden, D. and Shaw, G.A., "Hyperspectral image processing for automatic target
detection applications, Lincoln Laboratory Journal, Vol. 14, 79 (2003).

[7] Scharf, L. and Friedlander, B., "Matched subspace detectors," IEEE Transactions on Signal Processing,
Vol. 42, 2146 (1994).

[8] Fawcett, T., "An introduction to ROC analysis," Pattern Recognition Letters, Vol. 27, 861 (2006).

RTO-MP-IST-112 19 - 11
Airborne Infrared Hyperspectral Imager for
Intelligence, Surveillance and Reconnaissance Applications

19 - 12 RTO-MP-IST-112