The central wavelengths, Ao, of M-WIP filters 
ranging from 1797nm to 4512nm (at cryogenic 
temperature of 77K) correspond to the spectral sensitivity 
of the PtSi Schottky-barrier detectors (SBDs) used in the 
camera. The bandwidths of the filters at half of peak 
transmission range from 47 nm for the 1797-nm filter to 
102 nm for the 4512-nm filter. The values of the peaks of 
the filter transmissions were individually attenuated in 
order to provide approximately equal detected signal at all 
wavelengths for blackbody target at 700°C. The 
attenuation of the IR filters in this manner partially 
compensates for the spectral variations in the detected 
signal levels and, therefore, reduces the requirements on 
the dynamic range of the camera. 
Image Data Processing System 
Image data processing for M-WIP is based on the 
DATACUBE  MaxVideo system using a SUN 
workstation as the host computer. The DATACUBE 
MaxVideo system is a user configurable real-time image 
acquisition and processing system. It consists of a 
number of modules that can be interconnected to suit the 
image processing task. The modules are housed in a 20 
slot VME chassis box (MAX-BOX) while a SUN 4/330 
acts as the host computer. The system can operate at a 
maximum data rate of 10 MHz. In our system, we have 
one MAX-SCAN data acquisition module, two ROI- 
STORE 512Kbytes memory modules, three MAX-SP 
general purpose signal processing modules and one 
MAX-GRAPH video and graphics display module. A 
custom developed display board is used to reformat the 
320x122 image for standard RS-170 display. The 
development software for the MaxVideo system is 
ImageFlow which consists of a set of C-callable libraries. 
EXPERIMENTAL RESULTS 
Compensation for Inherent System Non-Linearities 
In order to obtain high radiometric accuracy of the 
M-WIP measurements, we have found it necessary to 
compensate for non-linearities of the IR imager response 
[3,4]. In the case of the PtSi IR-CCD imager used in the 
experimental M-WIP system, both the dark current and 
the spectral responsivity of the Schottky-barrier detectors 
(SBDs) depend on the SBD bias voltage and decrease 
with increasing level of integrated charge signal. This 
dependence of the dark current and responsivity on the 
detected signal is especially pronounced at high signal 
levels resulting in a "saturation-type" non-linearity. On 
the other hand, the reduction of the signal due to apparent 
charge trapping losses is especially evident at low signal 
levels and represents "offset-ty pe" non-linearity. 
In order to correct the detected signal for the dark 
current charge, the suppression of the dark current due to 
high signal level can be estimated according to the 
following algorithm: 
l. Using cold-shield to prevent radiative flux from 
reaching the detectors, the dark current charge is 
experimentally measured for a wide range of optical 
integration times [3,4]. The dark current is then 
approximated by an exponential function of 
integration time, tint, as: 
SER sure mg e Lat Ec (5) 
A derivative of Eq. (5) with respect to optical 
integration time, tint, represents the dark current 
which can be expressed as function of the 
accumulated signal: 
Qnod 
ot int 
d 
m b(Spe we sm C) m b(S te C) (6) 
Nye ( SE ad ) = I» ( S) = 
where S represents the total accumulated charge 
regardless of its source. 
2. The dark current component of the signal integrated 
by the imager viewing the radiant target can now be 
expressed as: 
line 
ST ) ea pc(s(r Lint ) dt = 
0 
= x trs ) FE ky vt ks) + b(kz = Olint 
(7) 
where S(T,t,) - kje?' «Kk, is the detected 
signal approximated by the exponential function of 
optical integration time, tint, and T is the 
temperature of the radiant target. 
Figure 3 illustrates the experimentally measured dark 
current charge, Sp md and the estimated dark current 
charge, SHamated corresponding to the imager viewing 
the blackbody target at 500°C, 600°C, 700°C, and 800°C 
through the 4512-nm filter. 
  
  
      
    
  
3000 
"I DARK CURRENT CHARGE 
= * Measured (with cold shield) - 1 17 
© 25000- Estimated for 4.5um filter 
o for signal levels @: 27] 
D * 500°C-2 37} 
© . se. 
M2 200004 600°C - 3 pia 
o * 700*C -4 
t " 800C-5 
o 150004 estimated 
i dc 
= 
Oo 
= 100004 
© 
e 
  
  
50004 M T 
0.066 0.116 
0166 ^ 0216 
Integration Time [s] 
Fig. 3. Measured and estimated dark current 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B1. Vienna 1996 
the ii 
repre 
bias: 
detec 
non- 
acco: 
Lee lc A TU ET Ne SG