EXPERIMENTAL MULTI-WAVELENGTH IMAGING PYROMETER FOR REMOTE SENSING 
OF TEMPERATURE PROFILES ON SURFACES WITH UNKNOWN EMISSIVITY 
Michael B. Kaplinsky, Jun Li, Nathaniel J. McCaffrey, Edwin S. H. Hou and Walter F. Kosonocky 
Electronic Imaging Center 
New Jersey Institute of Technology, Newark, New Jersey 07102-1982, USA 
KEY WORDS: Radiometric Temperature Measurement, Multispectral Radiometry, Infrared Sensor 
ABSTRACT 
Experimental Multi-wavelength Imaging Pyrometer (M-WIP) is presented for remote sensing of temperature profiles 
of targets with unknown spectrally varying emissivity. A software package was developed for calibration and real-time 
M-WIP measurements. An experimental 7-filter line-sensing M-WIP system was implemented with a 320x122-element 
PtSi IR-CCD imager and an assembly of narrow-band striped IR filters in the spectral range from 1797nm through 
4512nm. The M-WIP system was calibrated against a commercial blackbody source over temperature range from 450?C 
to 950°C. The signal processing included background subtraction, compensation for variation of dark current with 
detected signal and correction for non-linearities of IR imager response. Initial M-WIP measurements demonstrated real- 
time temperature resolution AT of +1°C for blackbody target over temperature range from 600°C to 900°C. Temperature 
resolution of +4°C was demonstrated for the blackbody source viewed through the double polished silicon wafer with 
unknown spectral transmissivity in the temperature range from 500°C to 950°C. 
INTRODUCTION In order to obtain high accuracy of M-WIP 
temperature measurements it is essential to compensate 
The Don-contact radiometric temperature for the inherent non-linearities of the IR imager response. 
measurements are based on the detection and analysis d The signal conditioning presented in this paper includes 
thermal e viotien emite by gn e ect dre Er compensation for the loss of sensor responsivity with 
à ihe blackbody radi 7 à is defined id T accumulated signal level (saturation), background 
Ot the blac Qdy, Iadiator, vic 15 GeHned as an ide subtraction, and correction for the accumulated dark 
surface that emits more thermal radiation than any other current charge 
surface at the same temperature. The emission of real ; 
objects can be described by the surface emissivity, &, 
which is defined as the ratio of radiation emitted by the 
real surface to that emitted by the blackbody at the same 
temperature. 
This paper also presents the least-squares based 
algorithm for calibration of M-WIP system against 
reference blackbody source. This calibration utilizes wide 
range of blackbody temperatures and yields effective 
; 3 spectral and transmission characteristics of the svstem. 
In general, in order to infer the temperature of the P S 
target from the measurement of emitted radiation, the 
value of the surface emissivity, e, should be known. 
Therefore, in the situations where the emissivity of te —. MULTI-WAVELENGTH IMAGING PYROMETER 
target is changing rapidly, or the conditions of the 
process preclude the independent measurements of target In its most general form, the least-squares 
emissivity, the conventional radiometric methods will minimization problem of M-WIP can be expressed as: 
not yield the true temperature. 
2 
: N ~ 
This paper presents the progress in the development min X A S;-S (TE À; | >. >. SA) (1) 
of a multi-wavelength imaging pyrometer (M-WIP) for TE i=l|0 
remote sensing of temperature profiles of hot surfaces with 
unknown wavelength-dependent emissivity. For remote where: 
temperature sensing by M-WIP, the spectral radiance of : 
the target is measured at several distinct wavelengths by Si — detected signal at A; (electrons / pixel), 
an IR-CCD camera with an assembly of 7 narrow-band S(T,£,A) — theoretical signal at A (electrons / pixel), 
filters positioned in front of its FPA. Based on these T — unknown temperature of the target (K), 
measurements the temperature and model parameters of €=€(A) — emissivity model, 
the target emissivity are determined simultaneously from a a e ; 6 
the least-squares fit of the theoretical model of IR camera Sos AS; = JSi Ims signal noises electronypixel) 
output signal to the experimental data [1-4, 11]. 
i 
78 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B1. Vienna 1996 
It 
calls 1 
differe 
values 
values 
U 
signal 
refere 
where 
RA) | 
T) is 
effecti 
1(À),t 
integr 
based 
on the 
deterr 
emiss 
comp 
accure 
On th 
will 
accur: 
of fre 
Analy 
of vai 
spectr 
polyn 
where 
mode 
E 
of th 
Trans 
apply 
indicz 
appro 
the e» 
with ¢ 
image