sensors

Article
Active Hyperspectral Sensor Based on MEMS
Fabry-Pérot Interferometer
Teemu Kääriäinen *, Priit Jaanson, Aigar Vaigu, Rami Mannila and Albert Manninen
VTT Technical Research Centre of Finland Ltd, Vuorimiehentie 3, 02150 Espoo, Finland; [email protected] (P.J.);
[email protected] (A.V.); [email protected] (R.M.); [email protected] (A.M.)
* Correspondence: [email protected]; Tel.: +35-85-0415-5981




Received: 31 March 2019; Accepted: 5 May 2019; Published: 12 May 2019


Abstract: An active hyperspectral sensor (AHS) was developed for target detection and classification
applications. AHS measures light scattered from a target, illuminated by a broadband near-infrared
supercontinuum (SC) light source. Spectral discrimination is based on a voltage-tunable MEMS
Fabry-Pérot Interferometer (FPI). The broadband light is filtered by the FPI prior to transmitting,
allowing for a high spectral-power density within the eye-safety limits. The approach also allows
for a cost-efficient correction of the SC instability, employing a non-dispersive reference detector.
A precision of 0.1% and long-term stability better than 0.5% were demonstrated in laboratory tests.
The prototype was mounted on a car for field measurements. Several road types and objects were
distinguished based on the spectral response of the sensor targeted in front of the car.

Keywords: hyperspectral; lidar; remote sensing; supercontinuum; FPI

1. Introduction
Hyperspectral remote sensing refers to a remote spectral detection of light, reflected or scattered
from a target. Each pixel of a hyperspectral imager can contain hundreds of spectral channels,
as opposed to the traditional three-color RGB cameras. Hyperspectral imaging has been widely applied
in applications such as medical imaging and diagnostics [1], food safety inspection [2] and agriculture
studies [3]. Hyperspectral cameras are usually dependent on ambient lighting. This limits the accuracy
of the spectral signal since any variation in the illumination spectrum translates into a misinterpretation
of the target response.
Active hyperspectral sensing refers to a method where the investigated target is artificially
illuminated by a broadband light source. A supercontinuum (SC) light source is most commonly used,
however optical parametric oscillators (OPO) [4], as well as frequency combs [5], have also been used.
SC is generated via spectral broadening of a short high-power monochromatic laser pulse in an optically
non-linear material, usually an optical fiber [6]. The resulting light can have an optical bandwidth
spanning over multiple octaves [7] but can still be directed and focused over long distances [8–11] as a
regular laser. The potential applications for the technology include mineral survey, the automotive
industry and agriculture.
The simplest case of an active hyperspectral sensor (AHS) or imager is a passive hyperspectral
imager combined with an external light source to provide more light at the measured wavelengths [12].
However, as with passive illumination, the spectral radiant intensity of the external light source can drift.
To extract the accurate spectral response of the target, a calibration target should be used at the target
scene. A more attractive way is to calibrate the response with a calibration target only once and actively
measure the spectrum of the outgoing beam for reference. Any drifts in the received signal caused by
external illumination can be compensated by normalizing the received signal with the reference signal.
This enables much more accurate measurements of the target spectral response compared to passive

Sensors 2019, 19, 2192; doi:10.3390/s19092192 www.mdpi.com/journal/sensors

Sensors 2019, 19, 2192 2 of 11

illumination, without the need for frequent calibrations of the spectral response. The active referencing
is important when using SC illumination. The SC output is highly dependent on any variation in the
fiber geometry and laser pulse properties. This usually results in large pulse-to-pulse variations and a
drift of the output optical power and spectrum.
The active monitoring of the outgoing beam can be realized with fast detection where the outgoing
and received pulse are recorded with a single element. This approach has high requirements for
the detection speed and cannot be used with integrating detectors. Nevertheless, fast detection will
allow for combining the ranging measurement with the spectral measurement [13–16]. An alternative
approach is to use two spectrometers, one monitoring the transmitted spectrum and another for the
received spectrum. This approach requires two separate components to disperse the light, which
results in a higher cost and complexity of the instrument. Thirdly, if the SC is already spectrally filtered
prior to transmission, separate single-color point detectors can be used to detect both transmitted and
received light. This approach poses a practical problem of implementing spectral scanning methods,
which do not result in wavelength dependent illumination patterns over long distances. This would be
the case if prisms or diffraction gratings were used for dispersion.
To date, relatively few prototypes of hyperspectral sensing using SC sources have been published
and no commercial instruments exists. More work remains in both instrumental design and application
studies in order to find the full commercial potential of the technology. In this work, we present an
AHS prototype, based on a voltage-tunable near-infrared MEMS Fabry-Pérot interferometer (FPI) [17].
The FPI allows for a compact, cost-efficient and fast on-axis wavelength selection between 1300 nm and
1650 nm. We used the FPI to select the illumination band of the in-house built SC source. This approach
enables the use of low-cost point detectors for both outgoing and received light. It also allows for
more light to be transmitted in eye-safety critical applications compared to illumination of the whole
bandwidth simultaneously. To demonstrate the approach, a compact battery-powered hyperspectral
sensor was built, characterized and tested in the field.
This article is organized as follows. In Section 2.1 we present the system design of the prototype.
Sections 2.2 and 2.3 present the SC generation and the FPI characterization, respectively. In Section 3.1
we present profiling of the outgoing beam after the FPI and transmission optics. Based on the results,
we extrapolate the spectral illumination pattern over long distances. In Section 3.2 we show the
results from Allan-Werle deviation analysis of the received pulse energy and demonstrate how long
period, calibration-free operation can be achieved with the active referencing. In Section 3.3 we present
the results from the field measurements. The prototype is mounted on a test car and we present
measurement results provided by the sensor, probing the road condition ahead of the moving vehicle.
Finally, in Section 4 we conclude the work.

2. Materials and Methods

2.1. Instrument Design

The optical design of the AHS sensor is shown in Figure 1a. The SC module and FPI are explained
in more detail in Sections 2.2 and 2.3, respectively. The SC output is coupled to the transmitter module
with a fiber connector. The wavelengths below 1300 nm are filtered out from the SC output with
a reflective optical long-pass filter. The wavelength band <1300 nm, reflected from the long-pass
filter, is measured by a fast InGaAs photodiode (Thorlabs FGA015, Newton, NJ, USA) connected to a
comparator circuit to provide a timing signal for the data-acquisition. Wavelengths >1300 nm enter
the FPI for spectral band selection. A small portion (1–5%) of the FPI-selected transmission band
is reflected by a beam sampler onto the second InGaAs photodiode (Thorlabs FGA10) to provide a
reference signal. The rest of the FPI-selected light is transmitted by the beam splitter. The transmitted
light is expanded to approximately 3 mm 1/e2 diameter and collimated with convex (M1) and concave
(M2) mirrors.
 Sensors2019,
Sensors 2019,19,
18,2192
x FOR PEER REVIEW 33ofof11
11

(a) (b)
Figure1. 1.
Figure Schematic
Schematic of theofactive
the hyperspectral
active hyperspectral
sensor (AHS) sensor (AHS)consisting
instrument instrument consisting of
of supercontinuum
supercontinuum
(SC), transmitter (SC),
and transmitter and receiver
receiver modules PL = pump
(a). modules (a). PLlaser,
= pump = supercontinuum
SCFlaser, SCF = supercontinuumfiber,
L1 = short
fiber, L1 = short
focal focal
length length
lens lens for fiber
for fiber coupling, L2 =
coupling, L2short
= short focal
focal lens
lens forforfocusing
focusinglight
lightthrough
through
theFabry-Pérot
the Fabry-Pérotinterferometer
interferometer(FPI), LPF==long
(FPI),LPF longpass
passfilter,
filter,TDTD==trigger
triggerdetector,
detector,RD RD==reference
reference
detector, M1-M2 =
detector, M1-M2 = convex
convexand andconcave
concavemirrors
mirrors forfor
beam
beam expansion,
expansion,L3 =L3 = light
light collecting lens, DM
collecting lens,=
DM = dichroic
dichroic mirror, SD = SD
mirror, = signal
signal detector, L4 = L4
detector, = focusing
focusing lens. lens.
CADCAD drawing of the
drawing of prototype (b). (b).
the prototype The
The transmitter module is shown
transmitter module is shown in blue. in blue.

In
Inoptical
opticalsystems
systemswith withlonglonglight-propagation
light-propagationdistances distancesand andwidewidewavelength
wavelengthbands,bands,careful
careful
design
designisisessential
essentialto toavoid
avoidexcessive
excessivechromatic
chromatic aberration-related
aberration-related spectral spectral dispersion
dispersion of of the
the light.
light.
Despite the drawback, a lens was used in the design to allow for axial
Despite the drawback, a lens was used in the design to allow for axial beam propagation through the beam propagation through
the small
small aperture
aperture of the
of the FPI.
FPI. The The effect
effect ononthethe beam
beam propagation
propagation was was characterized
characterized andand
is is discussed
discussed in
in results.
results.
The
Thecollimated
collimated light
light from
fromthe thetransmitter
transmitter module
module isis overlapped
overlapped with with the
the field
fieldofofview
viewof ofthe
the
receiver
receivermodule.
module.The Thereflected
reflectedlightlightisiscollected
collectedby byaaspherical
sphericalplano-convex
plano-convexlens lens(L3)
(L3)with
withaa75 75mmmm
diameter.
diameter. A A dichroic
dichroic mirror
mirror reflects
reflects visible
visible wavelengths
wavelengths onto onto aa CCD CCD camera
camera for forvisualizing
visualizing the the
measured target. The received infrared light is focused onto the third
measured target. The received infrared light is focused onto the third InGaAs photodiode (Femto OE- InGaAs photodiode (Femto
OE-300-IN-03-FST,
300-IN-03-FST, Berlin, Berlin, Germany).
Germany). TheThe detector
detector has hasa aprogrammable
programmablegain gainin in order
order toto account
account for for
varying target reflectance and
varying target reflectance and distance. distance.
Both
Boththe thesignal
signalandandthe thereference
referencedetectors
detectorswere werelimited
limitedin inbandwidth
bandwidthto to11MHz,
MHz,eveneventhough
though
the
thelengths
lengthsof ofthe
theactual
actualSC SCpulses
pulseswere werelesslessthan
than1.5 1.5ns.
ns.TheThereceived
receivedpulsespulseswere
werestretched
stretchedin inthe
the
time domain in order to be digitized with slower and more cost-efficient
time domain in order to be digitized with slower and more cost-efficient analog-to-digital converters analog-to-digital converters
(ADCs).
(ADCs).The Thesignal
signalwas
wasdigitized
digitizedwith withan anopen
opensoftware
softwareFPGA-based
FPGA-baseddevelopment
developmentboard board(Red
(RedPitaya
Pitaya
STEM
STEMlab lab125-14)
125-14)with
with twotwo 14-bit
14-bit125125MS/s ADCs.
MS/s ADCs. A custom
A custom C-program
C-program was used
was to remove
used the offset
to remove the
from the recorded waveforms and to integrate the pulse signals. Only
offset from the recorded waveforms and to integrate the pulse signals. Only the final integrated the final integrated values were
transmitted
values werethrough Ethernet
transmitted connection
through Ethernet to aconnection
laptop forto spectral
a laptop analysis. A second
for spectral multifunction
analysis. A second
data-acquisition card (NI myRIO) was used for general control of the
multifunction data-acquisition card (NI myRIO) was used for general control of the instrument instrument (laser, CCD, detector
(laser,
gain),
CCD, detector gain), as well as creating the FPI voltage ramp for the spectral scanning. Thewas
as well as creating the FPI voltage ramp for the spectral scanning. The FPI voltage signal FPI
synchronized
voltage signalwith wasthesynchronized
acquisition using withthe theTTL trigger from
acquisition thethe
using trigger
TTLphotodiode.
trigger from A buffer was
the trigger
used in the FPGA
photodiode. to account
A buffer was used forinthe
thedelay
FPGAbetween
to account reading
for thethe trigger
delay and the
between actualthe
reading pulse.
trigger Even
and
though the external
the actual pulse. Evenlaptop was used
though in this development
the external laptop was used work,inboththisof the data-acquisition
development work, both cardsofare
the
capable of real-time
data-acquisition standalone
cards are capable operation.
of real-time standalone operation.
The
Themechanical
mechanical designdesign of of the instrument is shown shown in inFigure
Figure1b. 1b.TheTheinstrument
instrumentwas wasenclosed
enclosed in
in cm××3030cm
a a4040cm cm××20 20 cm
cm (l(l ××w × × h) water-proof
water-proof seamless
seamless 3D-printed
3D-printed plasticplastichousing.
housing. The Thetotal
total
instrument
instrumentweightweightwaswas99kg. kg.AnAneasily
easilyreplaceable
replaceabletilted tiltedfront
frontglass
glasswaswasusedusedto toshield
shieldthe
thetransceiver
transceiver
optics.
optics.TheThemetallic
metallicbase plate
base can be
plate canusedbe with
usedadhesive
with adhesiveheating pads
heating for operation
pads for in cold conditions.
operation in cold
The power consumed
conditions. The power byconsumed
the AHS sensor by the during
AHS the normal
sensor during operation
the normalwas 5operation
A with a 12 V power
was 5 A with source.
a 12
V power source.
 Sensors 2019, 19, 2192 4 of 11
Sensors 2019, 18, x FOR PEER REVIEW 4 of 11

2.2.
2.2. Supercontinuum
Supercontinuum Generation
Generation
The
The SCSC waswas generated
generated withwith a 1064
a 1064 nm nm pulsed
pulsed microchip
microchip laser
laser (Horus
(Horus Laser
Laser HLX-I-F020-000,
HLX-I-F020-000,
Limoges, France). The laser emits 25 kW peak power, 1.2 ns
Limoges, France). The laser emits 25 kW peak power, 1.2 ns pulses at a 25 kHz repetition pulses at a 25 kHz repetition
rate. The rate.
The output
output from the from theis laser
laser coupled is coupled to a low-cost
to a low-cost multimode multimode graded-index
graded-index telecom
telecom fiber withfiber
a 50with
μm a
50 µm core diameter and a 1310 nm zero dispersion wavelength (Corning
core diameter and a 1310 nm zero dispersion wavelength (Corning InfiniCor50, New York, NY, USA). InfiniCor50, New York, NY,
The spectral power density of the resulting SC spectra, generated in different fiber lengths, are shownare
USA). The spectral power density of the resulting SC spectra, generated in different fiber lengths,
in shown
Figure 2. inThe
Figure 2. TheSC
resulting resulting
dependsSC ondepends on theofproperties
the properties the opticalof the optical
pump pulse as pump
well aspulse as well
the fiber.
as the fiber. Variation in coupling efficiency and any bending or movement
Variation in coupling efficiency and any bending or movement of the SC fiber have a big impact of the SC fiber haveon a
the resulting spectral power distribution. Moreover, the passive Q-switching of the pump laserthe
big impact on the resulting spectral power distribution. Moreover, the passive Q-switching of
pumpinlaser
results results
a high in a high
fluctuation of fluctuation
the pump laserof the pump
pulse laser nm.
at 1064 pulse at 1064
The nm. The SC
SC generation was generation
tested with was
tested with multiple fiber lengths to find the optimal spectral power distribution
multiple fiber lengths to find the optimal spectral power distribution in the measurement bandwidth in the measurement
bandwidth
between 1300between
nm and 1300 1700nm andnm. 1700
The nm. The spectral
spectral measurements
measurements were were conducted
conducted withwith a grating
a grating
spectrometer (OceanOptics NIRQuest 512-2.5, Seminole, FL, USA).
spectrometer (OceanOptics NIRQuest 512-2.5, Seminole, FL, USA). Based on the measurements Based on the measurements shown
in Figure 2, a fiber length of 100 m was used in the instrument. The SC
shown in Figure 2, a fiber length of 100 m was used in the instrument. The SC generation is more generation is more efficient
with longer
efficient fibers due
with longer fibers todue
an increased probability
to an increased for thefor
probability nonlinear effects.
the nonlinear However,
effects. if the iffiber
However, the is
too long, fiber absorption losses after 2250 nm start to decrease the output
fiber is too long, fiber absorption losses after 2250 nm start to decrease the output power. Too-short power. Too-short fiber
fiber results in an inefficient energy conversion to longer wavelengths. The compromise was made atm,
results in an inefficient energy conversion to longer wavelengths. The compromise was made at 100
100where the average
m, where radiant
the average powerpower
radiant withinwithin
the desired measurement
the desired bandwidth
measurement of 1300–1650
bandwidth nm was
of 1300–1650
approximately 40 mW of the 500 mW
nm was approximately 40 mW of the 500 mW pump power. pump power.

Figure 2. Spectral power density of the generated SC. The generation of longer wavelengths is more
Figure 2. Spectral power density of the generated SC. The generation of longer wavelengths is more
efficient with longer fibers, however, excessively long fiber results in a power-loss due to intrinsic fiber
efficient with longer fibers, however, excessively long fiber results in a power-loss due to intrinsic
absorption. Short fiber lengths result in an inefficient spectral broadening.
fiber absorption. Short fiber lengths result in an inefficient spectral broadening.
2.3. Fabry-PérotInterferometer
2.3. Fabry-PérotInterferometer
The FPI consist of two opposing reflecting surfaces separated by an air gap, acting as a Fabry-Pérot
The FPI[18].
resonator consist
Dueofto
two
theopposing reflecting
high reflectivity ofsurfaces separated
the mirrors, by an
only the airconstructively
light gap, acting as interfering
a Fabry-
Pérot resonator
between [18]. Due
the mirrors to the high
is transmitted. reflectivity
Thus, of the mirrors,
the gap between onlydetermines
the mirrors the light the
constructively
optical band.
interfering
The miniaturized FPI, developed by VTT, can be tailored for a wide range of wavelengthsthe
between the mirrors is transmitted. Thus, the gap between the mirrors determines and
optical band. The miniaturized FPI, developed by VTT, can be tailored for a wide
resolutions [17,19–21]. The gap and thus the wavelength band, is adjusted by applying voltage between range of
wavelengths and resolutions
two electrodes. The FPI used [17,19–21].
in this work The had
gapanand thus the
aperture wavelength
of 1.5 band, is adjusted
mm and operated by
with voltages
applying voltage between two electrodes. The FPI used in this work had an aperture
between 0–13 V, resulting in a central wavelength shift between 1300 nm and 1650 nm. The tuning of 1.5 mm and
operated with voltages
characteristics between
of the FPI 0–13 V,
are shown in resulting
Figure 3. inThea central wavelength
FPI wavelength andshift between
full-width at 1300 nm and
half maximum
1650 nm. The tuning characteristics of the FPI are shown in Figure 3. The FPI
(FWHM) were determined using a grating spectrometer (OceanOptics NIRQuest 512, Seminole, wavelength and full-FL,
width at half maximum (FWHM) were determined using a grating spectrometer
USA). The transmission data was fitted with a theoretical transmittance model of the Fabry-Pérot (OceanOptics
NIRQuest
cavity to 512, Seminole,
extract FL, wavelength
the central USA). The transmission
and FWHM. data was fittedroutine
An automated with a theoretical
characterizedtransmittance
the FPI at 100
model of the Fabry-Pérot cavity to extract the central wavelength and FWHM. An automated
voltage values for an accurate description of the response. The non-linear voltage-central wavelength routine
characterized the FPI at 100 voltage values for an accurate description of the response. The non-linear
voltage-central wavelength relationship of the FPI was fitted with a second-order polynomial
Sensors 2019, 19, 2192 5 of 11

Sensors 2019, 18, x FOR PEER REVIEW 5 of 11

relationship of the FPI was fitted with a second-order polynomial function. The fit was used to generate
afunction.
custom-form
The fitvoltage ramp
was used to resulting
generate in a linear wavelength
a custom-form voltagescan.
ramp resulting in a linear wavelength
scan.

(a) (b)
Figure 3. Tuning
Tuning characteristics of the FPI. (a) The
The central
central wavelength
wavelength and full-width at half maximum
at a given voltage. The red curve shows the relationship between FPI voltage and FWHM. The blue
curve shows the relationship between FPI voltage and the central bandpass
bandpass wavelength.
wavelength. (b) Spectral
transmission measured with a constant
constant voltage interval, indicated by differentcolors.
voltage interval, indicated by different colors.

3. Results
3. Results
3.1. Supercontinuum Propagation
3.1. Supercontinuum Propagation
In our previous work [8], the challenges of using refracting optics with the supercontinuum
In our previous work [8], the challenges of using refracting optics with the supercontinuum
illumination at longer distances (100 m and over) were recognised. In this work, a refracting lens
illumination at longer distances (100 m and over) were recognised. In this work, a refracting lens was
was needed after the supercontinuum to fit the beam through a small aperture of the FPI. In order
needed after the supercontinuum to fit the beam through a small aperture of the FPI. In order to
to estimate the spectral uniformity of the illumination at different target distances, the beam was
estimate the spectral uniformity of the illumination at different target distances, the beam was
profiled with a thermal camera (NIT Tachyon 6400, Madrid, Spain) after the transmitter module.
profiled with a thermal camera (NIT Tachyon 6400, Madrid, Spain) after the transmitter module. The
The beams spatial profile was measured at multiple distances between 0–12 m, scanning the central
beams spatial profile was measured at multiple distances between 0–12 m, scanning the central
wavelength at each distance with 50 nm increments. The measured spatial beam profile was fitted
wavelength at each distance with 50 nm increments. The measured spatial beam profile was fitted
with a two-dimensional Gaussian function. The beam profile resembled TEM00 spatial mode at all
with a two-dimensional Gaussian function. The beam profile resembled TEM00 spatial mode at all
wavelengths with no signs of higher-order spatial modes. The resulting maximum energy density of
wavelengths with no signs of higher-order spatial modes. The resulting maximum energy density of
the outgoing pulse was determined to be less than 0.8 µJ/cm22 at the wavelength of 1300 nm [22], which
the outgoing pulse was determined to be less than 0.8 μJ/cm at the wavelength of 1300 nm [22], which
is well below the eye-safety limits dictated by the Finnish Safety and Chemicals Agency (TUKES).
is well below the eye-safety limits dictated by the Finnish Safety and Chemicals Agency (TUKES).
The propagation of the beam was estimated by a Gaussian function. The results of the beam
The propagation of the beam was estimated by a Gaussian function. The results of the beam
profiling with a fit for Gaussian propagation are shown for a single wavelength in Figure 4a.
profiling with a fit for Gaussian propagation are shown for a single wavelength in Figure 4a. The
The ellipticity of the beam is the result of the astigmatism in the beam expander based on spherical
ellipticity of the beam is the result of the astigmatism in the beam expander based on spherical
mirrors. The average M22 value for both axes and all wavelengths was 4.0. The M22 value for the 8
mirrors. The average M value for both axes and all wavelengths was 4.0. The M value for the 8
wavelengths are shown in Figure 4b with the extrapolated 1/e22 diameter after 100 m of propagation.
wavelengths are shown in Figure 4b with the extrapolated 1/e diameter after 100 m of propagation.
As can be seen, the fitted M2 decreases at higher wavelengths. This behavior is likely due to the fact that
As can be seen, the fitted M2 decreases at higher wavelengths. This behavior is likely due to the fact
the SC generation is highly intensity-dependent. The power density threshold of the SC generation at
that the SC generation is highly intensity-dependent. The power density threshold of the SC
shorter wavelengths is reached over a broader spatial area of the fibre, whereas the cascaded threshold
generation at shorter wavelengths is reached over a broader spatial area of the fibre, whereas the
for generating longer wavelengths is only reached at the center of the fibre. These lower M2 values
cascaded threshold for generating longer wavelengths is only reached at the center of the fibre. These
compensate well for the larger divergences experienced by the longer wavelengths due to diffraction.
lower M2 values compensate well for the larger divergences experienced by the longer wavelengths
due to diffraction.
Sensors 2019, 19, 2192 6 of 11
Sensors 2019,
Sensors 2019, 18,
18, xx FOR
FOR PEER
PEER REVIEW
REVIEW 66 of
of 11
11

(a)
(a) (b)
(b)
Figure 4.
Figure 4. (a)
(a) Measured
Measured widths
widths forfor single
single (1.5
(1.5 µm)
μm) wavelength
μm) wavelength and
wavelength and aaa Gaussian
Gaussian fit
Gaussian fit Extrapolated
fit Extrapolated 1/e
Extrapolated 1/e222
1/e
diameter at
diameter at 100
100 m
m and
and (b)
(b) the
the fitted
fitted MM22 in
M
2 in function
function of
of wavelength.
wavelength.

3.2. Receiver Response

3.2. Receiver
Receiver Response
3.2. Response
The Allan-Werle deviation
deviation analysis [23] was
was used
used to to study
study the
the stability
stability andand noise
noise performance
performance
The Allan-Werle deviation analysis [23] was used to study the stability and noise performance
of the
the spectral
spectral measurement.
measurement. A white calibration target (Labsphere, (Labsphere, Spectralon
Spectralon SRS-99-050,
SRS-99-050, North
North
of the spectral measurement. A white calibration target (Labsphere, Spectralon SRS-99-050, North
Sutton, NH,NH,USA) USA)was wasplaced
placed 5 m from the instrument. The custom-form voltage ramp was
Sutton, NH, USA) was placed 55 m m from
from the the instrument.
instrument. The The custom-form
custom-form voltage
voltage rampramp was was applied
applied
applied
to FPI
FPI atattoaaFPI
30 at
Hza 30 Hz repetition
repetition rate. The rate.measurement
The The measurement was conducted
was conducted
conducted continuously
continuously 14for
for 14 h 14
in haa
to 30 Hz repetition rate. measurement was continuously for h in
in a temperature-controlled
temperature-controlled laboratory laboratory
laboratory space.
space. Anspace. An Allan-Werle
An Allan-Werle
Allan-Werle deviation deviation
deviation plot
plot of plot of
of aa single a single
single channel
channel (1500 channel
(1500 nmnm
temperature-controlled
(1500
± 2.5 nm ± 2.5 nm
nm center center wavelength)
wavelength) extracted extracted
from thefrom the spectra
spectra is shown is shown
in in Figure
Figure 5a. For5a. For comparison,
comparison, both
± 2.5 nm center wavelength) extracted from the spectra is shown in Figure 5a. For comparison, both
both
the rawthe raw
raw intensity
intensity andandthe the intensity
intensity corrected
corrected bybythethe referencedetector
reference detectorsignal
signalareare shown.
shown. The The noise
the intensity and the intensity corrected by the reference detector signal are shown. The noise
performance and stability of the measurement are considerably improved improved by the active active reference.
reference.
performance and stability of the measurement are considerably improved by the active reference.
Pulse-to-pulse
Pulse-to-pulse deviation
deviationisisisreduced
reducedfrom from 9%9%to to
2%2% andandthethe 7-hour Allan-Werle
7-hour Allan-Werle deviation is below
deviation 0.5%
is below
below
Pulse-to-pulse deviation reduced from 9% to 2% and the 7-hour Allan-Werle deviation is
for
0.5% allfor
wavelengths.
all The analysis
wavelengths. The was repeated
analysis was for the whole
repeated for wavelength
the whole range withrange
wavelength 10 nmwithincrements
10 nm
0.5% for all wavelengths. The analysis was repeated for the whole wavelength range with 10 nm
in the central
increments in wavelength.
in the
the central The
central wavelength.results
wavelength. The are shown
The results in
results are Figure
are shown
shown in 5b. The
in Figure1, 10
Figure 5b. and
5b. The 100
The 1,1, 10pulses,
10 and
and 100as well
100 pulses, as
pulses,
increments
the
as 7-h as
as well
well Allan-Werle
as the 7-hour deviation
the 7-hour Allan-Werle
Allan-Werle aredeviation
comparable
deviation with all wavelengths.
are comparable
are comparable with all
with The active reference
all wavelengths.
wavelengths. activeefficiently
The active
The reference
reference
replaces the otherwise
efficiently replaces
efficiently replaces the frequent
the otherwise need
otherwise frequent for
frequent needcalibrations
need for necessary
for calibrations to
calibrations necessaryaccount
necessary to for the
to account SC
account for drifts.
for the
the SC
SCThe best
drifts.
drifts.
possible
The best precision
The best possible of 0.1% isof
possible precision
precision achieved
of 0.1% is
0.1% with a 2000
is achieved
achieved with
withpulse average.
aa 2000
2000 pulse average.
pulse average.

(a)
(a) (b)
(b)
Figure
Figure 5.5. (a)
5. (a) Allan-Werle
Allan-Werle deviation
deviation of
of the measured pulse
the measured
measured pulse intensity
intensity with
with and
and without
without the reference
the reference
reference
signal.
signal. The center wavelength
wavelength of
of the
the channel
channel shown
shown is
is 1500
1500 ±
± 2.5 nm. Spectralon
Spectralon was used as the
signal. The center wavelength of the channel shown is 1500 ± 2.5 nm. Spectralon was used as the target target
at 5 m distance. The data was collected in 14 h. (b) The Allan-Werle deviation for averaged
at 5 m distance. The data was collected in 14 h. (b) The Allan-Werle deviation for averaged 1, 10 and1, 10 and
100 pulses
100
100 pulses as
pulses as well
as well as
well as 7-hour
7-hour values
values are
are shown
shown for wavelengths between
center wavelengths
for center wavelengths between 1350
between 1350 and
1350 and 1650
and 1650 nm with
nm with
10
10 nm
nm increments.
increments.
10 nm increments.

Using 4 can be achieved for

Using the
Using the programmable
the programmable gain
programmable gain of
gain of the
of the detector,
the detector,aaatotal
detector, totaldynamic
total dynamicrange
dynamic rangeof
range of~10
of ~1044 can
~10 can be
be achieved
achieved for
for
the
the AHS
AHS sensor.
sensor. The
The system
system response
response was
was measured
measured by
by varying
varying the
the signal
signal level,
level, both
both by
by obstructing
obstructing
the AHS sensor. The system response was measured by varying the signal level, both by obstructing
the field
the field of
of view
view and
and by
by changing
changing the
the programmable
programmable gain gain of
of the
the detector.
detector. AA total
total of
of 600,000
600,000 pulses
pulses
were collected. The measurement conditions in these measurements were the same as
were collected. The measurement conditions in these measurements were the same as in the Allan- in the Allan-
Sensors 2019, 19, 2192 7 of 11

the field of view and by changing the programmable gain of the detector. A total of 600,000 pulses 7were
Sensors 2019, 18, x FOR PEER REVIEW of 11
Sensors 2019, The
collected. measurement
18, x FOR PEER REVIEWconditions in these measurements were the same as in the Allan-Werle 7 of 11
deviation analysis.
Werle deviation Three different
analysis. gain levels
Three different gainwere
levelsused consecutively
were for five levels
used consecutively for fiveof levels
physical of
FOV
Werle
physicalobstruction.
deviation The total received
analysis. Three
FOV obstruction. power
different
The total levels
gain
received are plotted
levelslevels
power wereare in
used Figure 6a. With
consecutively
plotted in Figure for an ideal system,
five levels
6a. With an idealof
the intensity-normalized
physical
system, FOV
the obstruction.spectra
intensity-normalized should
The total
spectra all
received give
should thegive
power
all same
levels
theresult.
are
same The
plotted average
in The
result. Figurespectrum
6a. With
average ofanall
spectrum the
ideal of
measurements
system, was used as a reference
the intensity-normalized
all the meas andshould
spectra the spectra recorded
all give at different
the same result. conditions
The averagewere compared
spectrum of
to this
all reference. The 1σ standard deviation of the spectra is shown in Figure 6b. The 1σ standard
the meas
deviation was less than ±2.5% for all wavelengths.

(a) (b)
(a) (b)
Figure 6. (a) Total intensity measured with three different gain levels and 5 levels of signal
obstruction.
Figure (a)(b)
6.6. (a) Total Standard
1σintensity
Total deviation
intensity measured
measured of intensity-normalized
with with
three three different
different spectra.
gainand
gain levels levels andof5signal
5 levels levelsobstruction.
of signal
obstruction. (b) 1σ Standard deviation of intensity-normalized
(b) 1σ Standard deviation of intensity-normalized spectra. spectra.
In addition to the instrumental drifts, the atmospheric absorption of the signal differs at varying
In
conditionsaddition to the instrumental
and distances, instrumental
mainly due drifts, the atmospheric
to water absorption.absorption
One solutionof the signalthis
around differs
is toatlimit
varying
the
conditions
conditions
measurement distances,
andtodistances, mainly due to
mainly
water absorption-free waterwindows.
spectral absorption. Onework
One
In this solution
solution around
around
however, this
thethis is to
is
water limit the
absorption
measurement
band was included to water absorption-free
in the spectral band spectral windows.
since many In this work
interesting work however,
targetshowever, the water absorption
for the automotive industry,
band was
such as road included
included in the
in the
friction, spectral
takespectral band since
bandofsince
advantage many
thesemany interesting
interesting
wavelenghts. targets
targets
For for the automotive
for thespectral
long-range automotive industry,
industry,
measurement
such as road
with these friction,
friction, take
wavelengths, take advantage
advantage of
compensation forthese
of these wavelenghts.
wavelenghts.
the atmospheric For long-range
transmission spectralThis
is necessary. measurement
is however
with these wavelengths,
outside the scope of this work. compensation for the atmospheric transmission is necessary.
necessary. This is however
however
outside the scope of this work.
3.3. Field Trials
3.3. Field
3.3. Field Trials
Trials
To test the robustness of the sensor in field environment, the instrument was mounted on a roof
To
on test
rack To top the
test thearobustness
of test car, as of
robustness of the
the sensor
shown sensor
in Figurein field
in field environment,
illuminatedthe
environment,
7. The the instrument
instrument
detection was mounted
spot was mounted
directed at on10aa m
on roof
roof
in
rack
rack on
fronton top
of top of
the of a test car,
a test car,
vehicle. as shown
Theasshort
shown in Figure
in Figure
distance 7.
was7.usedThe illuminated
The illuminated detection
in this pilot detection spot was directed
spot was directed
study to optimize at 10 m
at 10 m in
the signal-to-noise in
front
front of
ratio. of
The the
the vehicle.
vehicle. The
instrument The short
wasshort distance
distance
powered was
by awas used 12
used
separate in V
in this
this pilot
carpilot study
study
battery. to optimize
to test
The optimize the
the signal-to-noise
run includedsignal-to-noise
an off-road
ratio.
ratio. The instrument
The
test track instrument
as was powered
was
well as urban powered by aaaseparate
by
road. With separate
lack of a12 12calibration
V car
V car battery.
battery. Thecomparisons
The
target, test run
test run included
included an
werean doneoff-road
off-road
with
test
test track
track
similar as well
as well
road as urban
as urban
surfaces, road.
mainly With
road. a lack
With at
asphalt, of
a lack a calibration
of a calibration
different target,
time pointstarget, comparisons were
comparisons
to confirm done with similar
were done drifts
that no significant with
road
similarsurfaces,
occurred. mainly asphalt,
road surfaces, mainly at different
asphalt, at time points
different timeto confirm
points to that no significant
confirm that no drifts occurred.
significant drifts
occurred.

Figure 7. The sensor mounted on the test vehicle.

Figure 7. The sensor mounted on the test vehicle.
The prototype operated flawlessly during the whole test event. Some chosen spectra of different
roadThe prototype
types with theoperated flawlessly
corresponding during
figure the whole
caption testintegrated
with the event. SomeCCDchosen spectra in
are shown of Figure
different
8.
road types with
The spectra theaverage
are an corresponding figure caption
of 10 individual spectrawith the integrated
measured CCD
at a 100 Hz rateare shown
during theindrive
Figure 8.
with
The spectra
speeds are an
between average
10–40 km/h.ofEach
10 individual
spectrum spectra measured
is normalized to a at a 100
mean Hz rate
value of 1.during the drive
The spectral with
response
speeds between 10–40 km/h. Each spectrum is normalized to a mean value of 1. The spectral response
Sensors 2019, 19, 2192 8 of 11

The prototype operated flawlessly during the whole test event. Some chosen spectra of different
road types with the corresponding figure caption with the integrated CCD are shown in Figure 8.
Sensors
The 2019, 18,
spectra x FOR
are PEER REVIEW
an average of 10 individual spectra measured at a 100 Hz rate during the drive 8with of 11
Sensors 2019, 18, x FOR PEER REVIEW
speeds between 10–40 km/h. Each spectrum is normalized to a mean value of 1. The spectral response 8 of 11
of the asphalt acted as the calibration surface and thus, is centered around one. In addition
of the asphalt acted as the calibration surface and thus, is centered around one. In addition to road to road
of the asphalt
condition, otheracted as thesuch
targets calibration surface
as cars, and thus,
pedestrians is centered
and obstaclesaround one. In addition spectral
to road
condition, other targets such as cars, pedestrians and obstacles producedproduced a distinct
a distinct spectral fingerprint.
condition,
fingerprint.other
Selectedtargets such
spectra as cars,
of objects arepedestrians and obstacles
shown in Figure 9. The speedproduced a distinct
of the vehicle spectral
was reduced
Selected spectra of objects are shown in Figure 9. The speed of the vehicle was reduced for object
fingerprint. Selected spectra
for object measurements, of objects
to only are shown
a few km/h. in Figure
The shown 9. The
spectra of speed of theand
a trash-can vehicle was reduced
a pedestrian were
measurements, to only a few km/h. The shown spectra of a trash-can and a pedestrian were measured
for object measurements,
measured with the car beingto only a few km/h. The shown spectra of a trash-can and a pedestrian were
stationary.
with the car being stationary.
measured with the car being stationary.

Figure 8. Selected spectra of road surfaces from the field-trial. Each spectrum is an average over 10
Figure 8. Selected spectra of road surfaces from the field-trial. Each spectrum is an average over
spectral8.measurements
Figure with
Selected spectra a 100surfaces
of road Hz measurement rate. The standard
from the field-trial. error of
Each spectrum theaverage
is an mean ofover
the 10
10 spectral measurements with a 100 Hz measurement rate. The standard error of the mean of
spectra are
spectral shown as error
measurements withbars.
a 100The
Hz test vehicle was
measurement driven
rate. The with a speed
standard errorranging between
of the mean 10–40
of the 10
the 10 spectra are shown as error bars. The test vehicle was driven with a speed ranging between
km/h. are shown as error bars. The test vehicle was driven with a speed ranging between 10–40
spectra
10–40 km/h.
km/h.

Figure 9. Selected spectra of objects from the field-trial. Each spectrum is an average over 10 spectral
Figure 9. Selected spectra of objects from the field-trial. Each spectrum is an average over 10 spectral
measurements with a 100 Hz measurement rate. The standard error of the mean of the 10 spectra are
measurements
Figure with
9. Selected a 100 of
spectra Hzobjects
measurement
from therate. The standard
field-trial. error of the
Each spectrum mean
is an of the
average 10 spectra
over are
10 spectral
shown as error bars. The test vehicle was driven with a speed ranging between 0–10 km/h.
shown as error with
measurements bars.aThe
100test
Hz vehicle was driven
measurement rate. with a speed ranging
The standard error of between
the mean0–10 km/h.
of the 10 spectra are
shown as error bars. The test vehicle was driven with a speed ranging between 0–10 km/h.
4. Discussions
4. Discussions
An AHS sensor based on MEMS FPI was built and demonstrated in both laboratory and in
4. Discussions
field.An
TheAHS
FPI sensor
enablesbased onon-axis
simple MEMS FPI was built
spectral andof
filtering demonstrated in both laboratory
the SC. For eye-safety and in field.
limited applications,
The An
FPI AHS sensor
enables based
simple on
on-axisMEMS FPI
spectral was built
filtering and
of thedemonstrated
SC. For in both
eye-safety laboratory
limited
the FPI-based sensor allows for greater transmitted spectral power density due to the spectral and in field.
applications, the
The FPI enables
FPI-based sensorsimple
allows on-axis spectral
for greater filteringspectral
transmitted of the SC. Fordensity
power eye-safety
duelimited applications,
to the spectral the
scanning
FPI-based sensor allows
of the transmitted lightfor greater transmitted
compared spectral
to transmitting thepower
wholedensity due to the spectral
SC. Furthermore, scanning
a simple active
of the transmitted
reference measurement lightusing
compared to transmitting
a single-point detection the
waswhole
shownSC. Furthermore,
to reduce the long-aand
simple active
short-term
reference measurement using a single-point detection was shown to reduce the long- and short-term
drifts considerably.
driftsThe
considerably.
SC source was characterized for precise directionality. The large observed M2 values of the
Sensors 2019, 19, 2192 9 of 11

scanning of the transmitted light compared to transmitting the whole SC. Furthermore, a simple active
reference measurement using a single-point detection was shown to reduce the long- and short-term
drifts considerably.
The SC source was characterized for precise directionality. The large observed M2 values of the SC
could be reduced by using single-mode fiber for SC generation. However, it seemed that the lower M2
of the longer wavelengths helped to uniform the illumination, correcting the difference in divergence
of different wavelengths due to diffraction. Also, the use of multimode fiber has practical advantages
on coupling the high-power pump light in to the fiber. The high peak-power pump pulses can damage
the fiber end if the light is focused to the cladding instead of the fiber core. The larger core of the
multimode fiber allows for more relaxed alignment.
The small low-cost MEMS FPI optical bandpass filter was used for wavelength selection. The FPI
filter does not refract different wavelengths and thus allows for light filtering before transmission of the
SC. The FPI did not show any signs of degradation or change in tuning properties due to high-intensity
laser light during this work. One of the limitations of using the FPI is the reduced bandwidth. The FPI
used in this work covered merely 25% of the whole SC spectrum. Future development of the FPI filter
might alleviate this limitation. Another practical limitation is the small aperture of the FPI that can
govern the transmitters optical design. Finally, with FPI used only at the transmission in the presented
prototype, the main detector is exposed to background radiation from the whole 350 nm FPI bandwidth.
For this reason, it might be advantageous to synchronize a second FPI to filter ambient light.
A best precision of 0.1% and long-term stability of 0.5% per spectral channel was demonstrated by
using the Allan-Werle analysis. This performance, however, is limited to stable atmospheric conditions
and received signal-strength. While this is not an issue in indoor applications, correction of the
atmospheric attenuation is critical in outdoor measurements due to changes in relative humidity.
Fog and dust particles also induce wavelength dependent scattering, affecting the spectral measurement.
With different detector gain levels and signal strengths, the overall 1σ standard deviation increased
to 2.5% for all wavelengths, a factor of 5 greater than with the stable signal. It is possible, that small
differences in the spectral uniformity affected the spectra measured with the light obstruction. More
work is required in order to asses the reasons behind the non-ideal behaviour.
The robustness of the instrument was put to test by mounting on a test vehicle. The sensor
acquired spectra of the road 10 m in front of the vehicle. Several road conditions and objects were
visually distinguished from the spectra. Future work should include building and integrating a model
for automated recognition of the road types and common objects. In the case of autonomous driving
or smart cars, the long-distance performance needs to be optimized. The sensor used in this work has
been tested for only relatively short distances. With the current prototype, stationary solid targets,
with reflectance larger than 0.1, have been measured with good SNR up to 50 m in lab conditions.
To increase the distance, avalanche photodiodes could be employed in the design.
With the addition of a scanner, the AHS could be fused as a part of a sensory platform in self
driving vehicles, providing target classification for objects, which are not recognized by conventional
methods, such as lidar or camera. Future work will also investigate the potential applications for the
sensor in other areas, such as mineral survey.

Author Contributions: T.K. and A.M. did the system design. R.M. designed the optics. P.J. and T.K. built the
instrument. T.K. and P.J. did the software. P.J. and T.K. conducted the measurements. T.K. and P.J. analyzed the
data. A.V. supported the activities from building the device to running the experiments. T.K. wrote the article.
A.M. managed and coordinated the research.
Funding: The research leading to these results has received funding from the European Commission and Business
Finland under ECSEL joint undertaking programme (DENSE project, Grant Agreement 692449) and Business
Finland Challenge Finland project Tehokas ja Turvallinen.
Acknowledgments: Ville Byman (c-program), Asko Rantanen (mechanical design) and Hans Toivanen (mechanical
design), Kari Ojasalo (power board design).
Conflicts of Interest: The authors declare no conflict of interest.
Sensors 2019, 19, 2192 10 of 11

References
1. Lu, G.; Fei, B. Medical hyperspectral imaging: A review. J. Biomed. Opt. 2014, 19, 010901. [CrossRef]
[PubMed]
2. Feng, Y.Z.; Sun, D.W. Application of hyperspectral imaging in food safety inspection and control: A review.
Crit. Rev. Food. Sci. 2012, 52, 1039–1058. [CrossRef] [PubMed]
3. Adão, T.; Hruška, J.; Pádua, L.; Bessa, J.; Peres, E.; Morais, R.; Sousa, J. Hyperspectral imaging: A review on
UAV-based sensors, data processing and applications for agriculture and forestry. Remote Sens. 2017, 9, 1110.
[CrossRef]
4. Howle, C.R.; Stothard, D.J.; Rae, C.F.; Ross, M.; Truscott, B.S.; Dyer, C.D.; Dunn, M.H. Active hyperspectral
imaging system for the detection of liquids. In Proceedings of the SPIE Defense and Security Symposium,
Orlando, FL, USA, 16–20 March 2008; Fountain, A.W., Patrick, J., Gardner, P.J., Eds.; SPIE: Washington, DC,
USA, 2008.
5. Boudreau, S.; Levasseur, S.; Perilla, C.; Roy, S.; Genest, J. Chemical detection with hyperspectral lidar using
dual frequency combs. Opt. Express 2013, 21, 7411–7418. [CrossRef] [PubMed]
6. Dudley, J.M. Supercontinuum Generation in Optical Fibers; Cambridge University Press: Cambridge, UK, 2010.
7. Silva, F.; Austin, D.R.; Thai, A.; Baudisch, M.; Hemmer, M.; Faccio, D.; Couairon, A.; Biegert, J. Multi-octave
supercontinuum generation from mid-infrared filamentation in a bulk crystal. Nat. Commun. 2012, 3, 807.
[CrossRef]
8. Manninen, A.; Kääriäinen, T.; Parviainen, T.; Buchter, S.; Heiliö, M.; Laurila, T. Long distance active
hyperspectral sensing using high-power near-infrared supercontinuum light source. Opt. Express 2014, 22,
7172–7177. [CrossRef] [PubMed]
9. Alexander, V.V.; Shi, Z.; Islam, M.N.; Ke, K.; Kalinchenko, G.; Freeman, M.J.; Ifarraguerri, A.; Meola, J.;
Absi, A.; Leonard, J.; et al. Field trial of active remote sensing using a high-power short-wave infrared
supercontinuum laser. Appl. Opt. 2013, 52, 6813–6823. [CrossRef] [PubMed]
10. Islam, M.N.; Freeman, M.J.; Peterson, L.M.; Ke, K.; Ifarraguerri, A.; Bailey, C.; Wager, M.; Anthony, A.;
Leonard, J.; Baker, H.; et al. Field tests for round-trip imaging at a 1.4 km distance with change detection
and ranging using a short-wave infrared super-continuum laser. Appl. Opt. 2016, 55, 1584–1602. [CrossRef]
[PubMed]
11. Orchard, D.A.; Turner, A.J.; Michaille, L.; Ridley, K.R. White light lasers for remote sensing. In Proceedings
of the SPIE Security + Defense, Wales, UK, 15–18 September 2008; David, H., Titterton, D.H., Mark, A.,
Richardson, M.A., Eds.; SPIE: Washington, DC, USA, 2008.
12. Nischan, M.; Joseph, R.; Libby, J. Active spectral imaging. Linc. Lab. 2003, 14, 131–144.
13. Hakala, T.; Suomalainen, J.; Kaasalainen, S.; Chen, Y. Full waveform hyperspectral LiDAR for terrestrial laser
scanning. Opt. Express 2012, 20, 7119–7127. [CrossRef] [PubMed]
14. Li, W.; Sun, G.; Niu, Z.; Gao, S.; Qiao, H. Estimation of leaf biochemical content using a novel hyperspectral
full-waveform LiDAR system. Remote Sens. Lett. 2014, 5, 693–702. [CrossRef]
15. Powers, M.A.; Davis, C.C. Spectral LADAR: Active range-resolved three-dimensional imaging spectroscopy.
Appl. Opt. 2012, 51, 1468–1478. [CrossRef] [PubMed]
16. Du, L.; Gong, W.; Shi, S.; Yang, J.; Sun, J.; Zhu, B.; Song, S. Estimation of rice leaf nitrogen contents based on
hyperspectral LIDAR. Int. J. Appl. Earth Obs. 2016, 44, 136–143. [CrossRef]
17. Rissanen, A.; Guo, B.; Saari, H.; Näsilä, A.; Mannila, R.; Akujärvi, A.; Ojanen, H. VTT’s Fabry-Perot
interferometer technologies for hyperspectral imaging and mobile sensing applications. In Proceedings of
the SPIE OPTO, San Francisco, CA, USA, 30 January–1 February 2017; Piyawattanametha, W., Park, Y., Eds.;
SPIE: Washington, DC, USA, 2017.
18. Perot, A.; Fabry, C. On the application of interference phenomena to the solution of various problems of
spectroscopy and metrology. Astrophys. J. 1899, 9, 87. [CrossRef]
19. Akujärvi, A.; Guo, B.; Mannila, R.; Rissanen, A. MOEMS FPI sensors for NIR-MIR microspectrometer
applications. In Proceedings of the SPIE OPTO, San Francisco, CA, USA, 13–18 February 2016;
Piyawattanametha, W., Park, Y., Eds.; SPIE: Washington, DC, USA, 2016.
20. Rissanen, A.; Akujärvi, A.; Antila, J.E.; Blomberg, M.; Saari, H.K. MOEMS miniature spectrometers using
tuneable Fabry-Perot interferometers. J. Micro Nanolithogr. MEMS MOEMS 2012, 11, 023003. [CrossRef]
Sensors 2019, 19, 2192 11 of 11

21. Rissanen, A.; Mannila, R.; Tuohiniemi, M.; Akujärvi, A.; Antila, J. Tunable MOEMS Fabry-Perot interferometer
for miniaturized spectral sensing in near-infrared. In Proceedings of the SPIE MOEMS-MEMS, San Francisco,
CA, USA, 3–6 February 2014; Piyawattanametha, W., Park, Y., Eds.; SPIE: Washington, DC, USA, 2014.
22. Jaanson, P.; Vaigu, A.; Kääriäinen, T.; Mannila, R.; Lehtomäki, V.; Manninen, A. A continuously tunable
NIR laser and its applications in material classification. In Proceedings of the SPIE Remote Sensing, Berlin,
Germany, 10–13 September 2018; Singh, U.N., Tzeremes, G.D., Eds.; SPIE: Washington, DC, USA, 2018.
23. Werle, P.O.; Mücke, R.; Slemr, F. The limits of signal averaging in atmospheric trace-gas monitoring by
tunable diode-laser absorption spectroscopy (TDLAS). Appl. Phys. 1993, 5, 131–139. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).