Sei sulla pagina 1di 7

Lock-in thermography using a cellphone attachment infrared camera

Marjan Razani, Artur Parkhimchyk, and Nima Tabatabaei

Citation: AIP Advances 8, 035305 (2018); doi: 10.1063/1.5021601


View online: https://doi.org/10.1063/1.5021601
View Table of Contents: http://aip.scitation.org/toc/adv/8/3
Published by the American Institute of Physics

Articles you may be interested in


Prediction of the filtrate particle size distribution from the pore size distribution in membrane filtration:
Numerical correlations from computer simulations
AIP Advances 8, 035308 (2018); 10.1063/1.5009568

The influence of ligand charge and length on the assembly of Brome mosaic virus derived virus-like particles
with magnetic core
AIP Advances 8, 035005 (2018); 10.1063/1.5011138

Modification of strain and 2DEG density induced by wafer bending of AlGaN/GaN heterostructure: Influence
of edges caused by processing
AIP Advances 8, 035318 (2018); 10.1063/1.5020149

Multilayered current-induced domain wall motion in Pt/Tb-Co/Ta/Tb-Co/Pt magnetic wire


AIP Advances 8, 025309 (2018); 10.1063/1.5017814

Mechanical stress relaxation in adhesively clamped carbon nanotube resonators


AIP Advances 8, 025118 (2018); 10.1063/1.5020704

RF power absorption by plasma of low pressure low power inductive discharge located in the external
magnetic field
AIP Advances 8, 035217 (2018); 10.1063/1.5023631
AIP ADVANCES 8, 035305 (2018)

Lock-in thermography using a cellphone attachment


infrared camera
Marjan Razani, Artur Parkhimchyk, and Nima Tabatabaeia
York University, Department of Mechanical Engineering, 4700 Keele Street, Toronto,
Ontario M3J 1P3, Canada
(Received 5 January 2018; accepted 25 February 2018; published online 6 March 2018)

Lock-in thermography (LIT) is a thermal-wave-based, non-destructive testing, tech-


nique which has been widely utilized in research settings for characterization and
evaluation of biological and industrial materials. However, despite promising research
outcomes, the wide spread adaptation of LIT in industry, and its commercializa-
tion, is hindered by the high cost of the infrared cameras used in the LIT setups.
In this paper, we report on the feasibility of using inexpensive cellphone attachment
infrared cameras for performing LIT. While the cost of such cameras is over two
orders of magnitude less than their research-grade counterparts, our experimental
results on block sample with subsurface defects and tooth with early dental caries
suggest that acceptable performance can be achieved through careful instrumen-
tation and implementation of proper data acquisition and image processing steps.
We anticipate this study to pave the way for development of low-cost thermogra-
phy systems and their commercialization as inexpensive tools for non-destructive
testing of industrial samples as well as affordable clinical devices for diagnostic
imaging of biological tissues. © 2018 Author(s). All article content, except where
otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license
(http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1063/1.5021601

Lock-in thermography (LIT) is a non-destructive testing (NDT) technique that is widely used
in research settings for characterization and evaluation of materials. The idea behind LIT is to apply
intensity-modulated external excitation to a sample to create a modulated temperature field (i.e.,
a thermal-wave field) inside the sample and then study the temporal evolution of temperature on
sample surface with an infrared camera. In this arrangement, presence of subsurface defects and/or
inhomogeneities alters the local centroid of the thermal wave field and thus yields surface temporal
temperature evolutions with different amplitude and phase compared to those registered in intact
areas.1 LIT, compared to most other NDT techniques, has the advantages of being non-contact,
having the ability to inspect opaque1–6 and turbid materials,7–9 and being scalable (e.g., microscopic
LIT of leakages in integrated circuits2 vs. inspection of airplane parts6 ). Moreover, depending on
the application, different types of external excitation, such as optical,8 electrical,1,2,10 magnetic,3
mechanical waves (ultrasound)5 or even cyclic stress/strain,4 can be utilized to induce the thermal
wave field inside the sample. Due to these promising abilities, LIT has been successfully implemented
in research for defect detection and evaluation of broad range of materials, spanning from composite
materials,10 to semiconductor,11 to biological hard tissues.7–9,12 However, the wide spread adaptation
of LIT in industry, and its commercialization, has been hindered by the high cost of the system. This
paper aims to resolve this obstacle by providing a solution that can reduce the system cost by over
two orders of magnitude.
The major cost determinant of a LIT system is the infrared camera used for non-contact detection
of surface temperature evolution. In existing industry standard, infrared cameras are offered in three
distinct spectral ranges: Short- Wavelength InfraRed (SWIR: 1.4–3 µm), Mid- Wavelength InfraRed
(MWIR: 3–8 µm), and Long- Wavelength InfraRed (LWIR: 8–15 µm). While the MWIR and LWIR

a
Address all correspondence to: nima.tabatabaei@lassonde.yorku.ca

2158-3226/2018/8(3)/035305/6 8, 035305-1 © Author(s) 2018


035305-2 Razani, Parkhimchyk, and Tabatabaei AIP Advances 8, 035305 (2018)

cameras are widely used in thermography, SWIR cameras have limited thermography applications due
to the minimal thermal radiation in this spectral range. The sensor core of research grade MWIR and
LWIR cameras are usually made from Indium Antimonide (InSb) and Mercury Cadmium Telluride
(MCT), respectively, and are actively cooled with an expensive and bulky internal sterling cooling
mechanism to achieve minimal noise equivalent temperature difference (NETD usually better than
25mK). Less expensive infrared cameras with passively cooled microbolometer sensors (spectral
range: 8-14 µm) are also available. These cameras are priced significantly less than their actively
cooled research-grade counterparts (∼$10k vs $100k), but have inferior performance (NETD of
∼50mK and lower frame rates). Recently, in an effort to make infrared cameras more accessible to
the general public, manufacturers have started offering cellphone attachment infrared cameras that
cost significantly less than any other infrared camera offered to date. While the cost of these cameras
can be as low as $200, their NETD can be as poor as 150mK and frame rates as low as 9Hz. In this
study, we have characterized the performance of these inexpensive cellphone attachment infrared
cameras and have demonstrated the feasibility of using them for lock-in thermography investigation
of biological and non-biological samples.
The experimental setup, Fig. 1(a), consists of a fiber-coupled (200 µm core diameter) con-
tinuous wave near infrared laser with center wavelength of 808 nm (Jenoptik, Jena, Germany). A
laser controller unit is used to thermally stabilize the laser as well as to modulate its intensity. In
order to achieve uniform laser beam intensity over the interrogated region of interest of the sam-
ple, a collimator-optical diffuser system (Thorlabs, Newton, New Jersey, USA, F220SMA-780 and
ED1-C20-MD) is placed in front of the laser beam. To register the thermal responses of samples,
a low-cost (USD $500) cellphone attachment infrared camera (Compact Pro from Seek Thermal)
with a spectral range of 7.5 to 14 µm is focused on sample surface. The passively cooled camera
has a micro bolometer thermal sensor with resolution of 320 × 240 consisting of 76,800 pixels and
a maximum nominal frame rate of greater than 15 frames per second (fps). The camera connection
to computer is a standard USB interface. In order to obtain camera frames in a controlled manner,
we design a data acquisition program in C++ platform which acquires camera frames in a free-run
mode while registering the absolute time at which each frame is captured. This timing informa-
tion is used in the signal processing step to compensate for the systemic data acquisition delays
of the inexpensive camera readout electronics. To gauge the performance of our inexpensive sys-
tem we conducted lock-in thermography experiments at laser modulation frequencies of 1Hz, 2Hz,
and 5Hz.
Figure 1(b) depicts the signal conditioning/processing block diagram which was applied on the
time signals corresponding to each pixel (i.e., pixel value vs. time). Briefly, after acquisition of frames,
the effect of bulk heating was removed from the dataset through polynomial fitting of order seven. One
of the major challenges of performing lock-in thermography with inexpensive cellphone attachment
cameras is the non-uniform acquisition of frames (i.e., non-constant frame rate) which scrambles the
amplitude and phase information. Our experience suggests that, in general, two factors contribute to
the non-uniformity of data acquisition. First, the small variations in the acquisition period (the small
changes in slopes of line segments in Figs. 2(a) and 2(b)) and, second, the camera’s native Flat Field

FIG. 1. Schematic diagrams of (a) the thermophotonic imaging experimental setup and (b) the signal processing block diagram
applied to the time signal of each pixel.
035305-3 Razani, Parkhimchyk, and Tabatabaei AIP Advances 8, 035305 (2018)

FIG. 2. Acquisition timing information for 500 consecutive frames (a) before and (b) after stabilization of camera’s native flat
field correction procedure. Example time signals for a single pixel from the block sample over the shallow hole experiencing
modulated laser excitation (c) before and (d) after interpolation. (e) and (f) are signal spectrums of (c) and (d), respectively.
Reconstructed phase images, (g) without and (h) with, interpolation. Reconstructed phase images, (i) without and (j) with,
interpolation and application of system transfer function.

Correction procedure (FFC; also known as non-uniformity correction) which kicks in regularly and
disrupts frame acquisition (arrows in Figs. 2(a) and 2(b)). The plots of Figures 2(a) and 2(b) show that
the FFC process took 0.6±0.1 seconds to be completed, however, the time between FFC executions
increases as imaging session time increases. For the camera used in this study, the time between
FFC executions was stabilized after frame number of 4825±10 (or after about 4 minutes). In order
to compensate for non-uniform data sampling, we use the timing information that we registered for
each frame to perform spline interpolation at a uniform sampling frequency of 20Hz, Figs. 2(c) vs.
2(d). Once the data was resampled, fast Fourier transform (FFT) is applied to the signals. The signal
spectrum plots, Figs. 2(e) and 2(f), clearly demonstrate the effect of interpolation and resampling on
reconstruction of proper signal spectrum and enhancement of signal-to-noise ratio. Once the signals
are uniformly resampled, the magnitude and phase of the complex number corresponding to the
laser modulation frequency (i.e., the peak of signal spectrum) is used to find the absolute values
of amplitude and phase images for each pixel. Comparison of Figures 2(g) and 2(h) indicates that
resampling the signals at a uniform rate results in retrieving the phase information as the underlying
defect vaguely appears in Figure 2(h) but not in Figure 2(g). Finally, the system’s transfer function
was calculated and applied to the amplitude and phase images to correct for imperfections in the
acquisition electronics (e.g., delays caused by rolling shutter) as well as illumination and collection
optical systems. Comparison of Figures 2(i) and 2(j) clearly demonstrates the significance of uniform
data sampling on proper calculation of phase images as the subsurface defect can only be resolved in
the phase image obtained from the uniformly sampled data. The transfer function was calculated by
performing lock-in imaging on a thick Aluminum block which was painted black on the front surface,
resembling a semi-infinite blackbody sample, and subsequently fitting the data to the theoretical
responses and finding the compensation factors for each pixel.
To demonstrate the performance and clinical impact of the developed inexpensive setup, we
conducted experiments on two sets of samples. While the intension of the first set of experiment
was to demonstrate the accuracy of performance of the developed inexpensive lock-in thermogra-
phy system and its accordance to the fundamental basics of the thermal-wave science, the second
set of experiment was carried out on a biological sample to highlight the clinical relevance and
035305-4 Razani, Parkhimchyk, and Tabatabaei AIP Advances 8, 035305 (2018)

potential clinical impact of the developed technology. The first set of experiment was carried out on
an Aluminum block with two blind holes with diameters of 5 mm, simulating subsurface defects.
The remaining wall thicknesses of the deep and shallow holes were 200 µm and 500 µm, respec-
tively. The interrogated side of the sample was painted with a matt black paint to enhance optical
absorption on the surface. The second set of experiment was conducted on an extracted human tooth
which was obtained from local oral surgeons in accordance with the bio- and laser safety regulation
of the York University. After sample cleaning, a demineralization protocol using an acidic gel was
applied on a rectangular treatment window for 10 days to induce local demineralization. This artifi-
cial demineralization protocol is known for creating early dental caries in a controlled environment
by mimicking the cyclic demineralization and remineralization processes that take place in the oral
cavity in presence of dental plaques.13 The details of the controlled demineralization process can be
found in our previous publications.7–9
Figure 3 depicts the amplitude and phase images obtained from the block sample with two blind
holes. The block with blind holes sample design in normally used in the field of photothermal science
to gauge the performance of the systems by demonstrating the characteristic depth profilometry
nature 14,15 That is, in accordance with the concept of thermal diffusion length,
p of thermal waves.
µ = α/πfm ; where α is medium thermal diffusivity, one can control the effective inspection depth
by choosing the appropriate laser intensity modulation frequency (f m ). Consequently, conducting
lock-in thermography at low modulation frequencies enables one to see deeper into the sample, albeit
at the cost of inferior depth resolution, while experiments carried out at high modulation frequencies
selectively reveal information from close-to-surface defects. To ensure validity of comparison of
images, every image in Figure 3 has been normalized by its corresponding average value within a
semi-infinite/defect-free reference area (the red square schematically shown in the 1Hz amplitude
image of the deep hole). As such, these images depict the amplitude and phase values differences from
those of a semi-infinite/defect-free region (i.e., 1 and 0 for amplitude and phase, respectively). The
amplitude images obtained from the deep hole, located approximately 200 µm below the interrogation
surface, clearly detect the presence of defect at all modulation frequencies. As expected, at higher
modulation frequencies the amplitude values become smaller at the defective region because thermal
waves are more damped as they propagate into the sample.16 Moreover, due to the shallow depth of
defect, the drill bit pattern at bottom of the hole is resolved (arrow in Fig. 3). The amplitude images
taken from the shallow hole, located approximately 500 µm below the interrogation surface, can
resolve the defect at modulation frequencies of 1Hz and 2Hz. However, the defect cannot be reliably
detected at high modulation frequency of 5Hz as at this frequency the thermal diffusion length is

FIG. 3. (a) Amplitude and (b) Phase images of deep hole (shallow defect; wall thickness ∼200µm) and shallow hole (deep
defect; wall thickness ∼500µm) at several laser modulation frequencies. All images have been normalized with respect to the
mean value in the semi-infinite reference area (red square) depicted in the top left image. The units for amplitude and phase
values are a.u. and radians, respectively.
035305-5 Razani, Parkhimchyk, and Tabatabaei AIP Advances 8, 035305 (2018)

not large enough to allow for detection of the deep feature. The phase images obtained from shallow
and deep holes are also consistent with the basic principles of thermal waves. Both holes can be
detected at all frequencies. The imaging resolution improves as modulation frequency is increased
due to smaller dispersion of high frequency thermal waves.16 Moreover, it can be seen that, unlike
the amplitude channel, the phase channel can resolve the shallow hole (i.e., deep defect) even at the
high modulation frequency of 5Hz. This observation is consistent with prior reports in the literature
and is due to the well-known emissivity-normalized nature17,18 of phase channel which makes it
more sensitive, compared to the amplitude channel, for detection of deep sub surface features.19
Finally, as expected, both amplitude and phase values tend toward those of the semi-infinite/defect-
free region at any given modulation frequency as the defect depth increases; thus, reducing the defect
contrast in images. The lock-in amplitude and phase images of Figure 3 demonstrate the alignment
of the results obtained from the developed inexpensive system with those previously obtained with
expensive research-grade infrared cameras.14
In recent years, lock-in thermography has proven to be effective for early detection of dental
caries.7–9,12 However, despite promising results, lock-in thermography has not yet been translated
to Dentistry as a commercial product mostly due to the high cost of the system (costing about USD
$120k with cryogenically cooled infrared cameras). Our next set of experiments aims to demonstrate
the possibility of translating the lock-in thermography technology to Dentistry using an infrared
camera that only costs USD $500.
Figure 4 shows the results obtained with our inexpensive system from an artificially induced
early caries. The treated rectangular window can clearly be observed in both LIT amplitude and
phase images while the visual image shows no indication of mineral loss. The treatment window
is seen as an area of higher intensity in the amplitude image because the demineralization process
enhances both the local absorption and scattering coefficients, resulting in more efficient absorption
of light and thus generation of thermal waves with larger amplitudes.20 Moreover, the local trapping
and absorption of light within the close-to-surface demineralized area shifts the local centroid of
the thermal wave field and causes the window to appear with a different contrast compared to the
surrounding intact areas in the phase image. The results of Figure 4 are in full alignment with our
previous LIT studies of early dental caries with high-end research-grade infrared cameras7,8 and,
thus, demonstrate feasibility of detecting early dental caries using cellphone attachment infrared
cameras.
We have demonstrated, for the first time, a lock-in thermography imaging system that incorporates
inexpensive cellphone attachment infrared cameras. The camera used in this study (Compact Pro; Seek
Thermal) was originally designed for general purpose thermography applications using cellphones;
however, through implementation of careful data acquisition and signal processing schemes, we
demonstrated the possibility of utilizing this inexpensive camera in a scientific setting for non-
destructive evaluation of biological and non-biological materials. Through experiments conducted
on a standard block sample with blind holes we demonstrated alignment of our inexpensive system
performance with basic principles of thermal wave science as well as its ability in detecting subsurface
defects in opaque materials. Our results from a tooth sample with an artificially induced caries show
the potential of the proposed system for early detection of dental caries. The cost of the camera
used in this study is about 200 times less than that of the cryogenic mid infrared cameras used
in the conventional scientific thermography settings ($500 vs. ∼$100,000), yet our results suggests
acceptable performance. Therefore, we anticipate this study to pave the way for development of

FIG. 4. (a) Visual and lock-in thermography (b) amplitude and (c) phase images of tooth with artificially-induced caries of
10 days within a rectangular treatment window. LIT images were obtained at laser modulation frequency of 2 Hz.
035305-6 Razani, Parkhimchyk, and Tabatabaei AIP Advances 8, 035305 (2018)

affordable and commercially-viable thermography systems which utilize pulsed or continuous wave
excitations for non-destructive evaluation of industrial samples and manufactured parts as well as
biological tissues such as tooth.
N.T. is grateful to the Natural Sciences and Engineering Research Council of Canada for the
award of Discovery Grant (RGPIN-2015-03666), to the Canadian Institutes of Health Research for
the award of Collaborative Health Research Projects grant (DAN381313), and to the Lassonde School
of Engineering and the York University for their financial support.
1 O. Breitenstein and M. Langenkamp, Lock-in Thermography (Springer, Berlin, 2003).
2 O. Breitenstein, M. Langenkamp, F. Altmann, D. Katzer, A. Lindner, and H. Eggers, “Microscopic lock-in thermography
investigation of leakage sites in integrated circuits,” Review of Scientific Instruments 71(11), 4155–4160 (2000).
3 P. Jäckel and U. Netzelmann, “The influence of external magnetic fields on crack contrast in magnetic steel detected by

induction thermography,” Quantitative InfraRed Thermography Journal 10(2), 237–247 (2013).


4 L. Krstulovic-Opara, B. Klarin, P. Neves, and Z. Domazet, “Thermal imaging and thermoelastic stress analysis of impact

damage of composite materials,” Engineering Failure Analysis 18(2), 713–719 (2011).


5 A. Mendioroz, R. Celorrio, and A. Salazar, “Ultrasound excited thermography: An efficient tool for the characterization of

vertical cracks,” Measurement Science and Technology 28(11) (2017).


6 D. Wu, A. Salerno, U. Malter, R. Aoki, R. Kochendörfer, P. Kächele, K. Woithe, K. Pfister, and G. Busse, “Inspection

of aircraft structural components using lockin-thermography,” Quantitative Infrared Thermography, QIRT 96, 251–256
(1996).
7 A. Ojaghi, A. Parkhimchyk, and N. Tabatabaei, “First step toward translation of thermophotonic lock-in imaging to dentistry

as an early caries detection technology,” Journal of Biomedical Optics 21(9), 096003 (2016).
8 N. Tabatabaei, A. Mandelis, and B. T. Amaechi, “Thermophotonic lock-in imaging of early demineralized and carious

lesions in human teeth,” Journal of Biomedical Optics 16(7), 071402-1–071402-10 (2011).


9 N. Tabatabaei, A. Mandelis, M. Dehghany, K. H. Michaelian, and B. T. Amaechi, “On the sensitivity of thermophotonic

lock-in imaging and polarized Raman spectroscopy to early dental caries diagnosis,” Journal of Biomedical Optics 17(2),
0250021–0250025 (2012).
10 R. Yang and Y. He, “Optically and non-optically excited thermography for composites: A review,” Infrared Physics and

Technology 75, 26–50 (2016).


11 O. Breitenstein, J. P. Rakotoniaina, M. H. Al Rifai, and M. Werner, “Shunt types in crystalline silicon solar cells,” Progress

in Photovoltaics: Research and Applications 12(7), 529–538 (2004).


12 N. Tabatabaei and A. Mandelis, “Thermal coherence tomography: Depth-resolved imaging in parabolic diffusion-wave

fields using the thermal-wave radar,” International Journal of Thermophysics 33(10-11), 1989–1995 (2012).
13 A. Hellen, A. Mandelis, Y. Finer, and B. T. Amaechi, “Quantitative evaluation of the kinetics of human enamel simulated

caries using photothermal radiometry and modulated luminescence,” Journal of Biomedical Optics 16(7), 071406 (2011).
14 N. Tabatabaei and A. Mandelis, “Thermal-wave radar: A novel subsurface imaging modality with extended depth-resolution

dynamic range,” Review of Scientific Instruments 80(3), 034902 (2009).


15 R. Mulaveesala and S. Venkata Ghali, “Coded excitation for infrared non-destructive testing of carbon fiber reinforced

plastics,” The Review of Scientific Instruments 82(5), 054902 (2011).


16 A. Mandelis, Diffusion-Wave Fields: Mathematical Methods and Green Functions (Springer, New York, 2013).
17 A. Rosencwaig and G. Busse, “High-resolution photoacoustic thermal-wave microscopy,” Applied Physics Letters 36(9),

725–727 (1980).
18 G. Busse and A. Ograbeck, “Optoacoustic images,” Journal of Applied Physics 51(7), 3576–3578 (1980).
19 G. Busse, D. Wu, and W. Karpen, “Thermal wave imaging with phase sensitive modulated thermography,” Journal of

Applied Physics 71(8), 3962–3965 (1992).


20 R. J. Jeon, A. Hellen, A. Matvienko, A. Mandelis, S. H. Abrams, and B. T. Amaechi, “In vitro detection and quantification

of enamel and root caries using infrared photothermal radiometry and modulated luminescence,” Journal of Biomedical
Optics 13(3), 034025 (2008).

Potrebbero piacerti anche