Analytical models of thermal imaging system performance have gradually become obsolete as TIS (thermal imaging systems) have become more complex and have improved in vertical performance. In particular, the effects of sampling and aliasing have not been included directly, but have had to be accounted for by side calculations before entering the data. In this paper, an approach to modeling second generation TIS is described in which the effects of sampling on both signal and noise are accounted for without requiring the user to make subsidiary calculations. The model is two-dimensional, using both vertical and horizontal resolution in the prediction of recognition and detection performance. A model for human perception is presented which differs slightly from the matched filter concept models and gives a closer match to measured data. The differences between modeling scanning and staring systems is discussed, as well as between systems with on-focal- rather than off-focal-plane sampling. Proper treatment of the several sources of noise in sampled systems is analyzed, including aliased noise.
The Night Vision Laboratory static performance model is considered for thermal viewing systems. Since the model is not initially intended to be a design tool and is not usable for conducting system or component design trades, it has to be restructured. The approach to updating the first-generation static performance model and to configuring it as a design tool is presented. Second-generation imaging systems exploit infrared focal-plane arrays, high-reliability cryogenic coolers, precision scanning devices, and high-speed digital electronics. They also use optical materials and coatings and optomechanical and electronics packaging techniques.
A model for staring focal plane arrays is described. The model characterizes two types of detectors, the platinum silicide (PtSi) Schottky-barrier diode and the mercury-cadmiun-tellurium (MCT) detector. The MCT devices provide high performance in both the 3-5-micron and 8-12-micron regions, while the Schottky devices, functioning only in the 3-5-micron band, provide high density and economy.
Performance improvement of a dual-band (3-5-micron and 8-12-micron) imaging radiometer is considered. The results of optical and subsystem modulation transfer function (MTF) analyses indicate that primary performance limitations include severe chromatic aberration in the 3-5-micron (SW) channel and the combination of slow SW detector time constant and poorly tuned high-frequency boost circuit. A new SW detector lens and telescope objective pair have been designed to reduce the chromatic aberration. An improved boost circuit provides MTF with the same noise-equivalent bandwidth. An InSb detector/preamp hybrid is being investigated for a possible replacement of the low-bandwidth HgCdTePC detector currently being used.
The end-to-end performance of image gathering, coding, and restoration as a whole is considered. This approach is based on the pivotal relationship that exists between the spectral information density of the transmitted signal and the restorability of images from this signal. The information-theoretical assessment accounts for (1) the information density and efficiency of the acquired signal as a function of the image-gathering system design and the radiance-field statistics, and (2) the improvement in information efficiency and data compression that can be gained by combining image gathering with coding to reduce the signal redundancy and irrelevancy. It is concluded that images can be restored with better quality and from fewer data as the information efficiency of the data is increased. The restoration correctly explains the image gathering and coding processes and effectively suppresses the image-display degradations.
In its current forni, the NVL model does not adequately predict the
laboratory measured minimum resolvable temperature (MRT) values at low
or high spatial frequencies. The differences between the measured and
the predicted values are caused by inappropriate modeling of the eye,
tremendous variability in observers, and ill-defined data analysis
methodology. In the usual laboratory procedure, the observer is allowed
to move his head. This, in effect, renioves the eye's response from the
model because by adjusting his viewing distance, the observer appears to
achieve equal detection capability at all spatial frequencies. Recent
studies provide two new eye models: one allowing head movement and one
in which the head is stationary.
A NVL type model was modified to accept five different eye models:
the original NVL eye model, the Sendall-Rosell model, the two new eye
models, and the Campbell-Robson eye model. All the models provided the
same shaped MRT curve at high spatial frequencies to within a constant.
Large discrepancies exist at low spatial frequencies presumably due to
the inability to model the eye's inhibitory response.
A series of experiments are described which quantify an operator's ability to detect
an unresolved target in backgrounds representative of a slow scan, large volume
infrared surveillance system. An inverse relationship was found between screen
refresh rates and the probability of target detection, with high refresh rates
asymptotically approaching a contrast limited threshold and low rates producing a
higher detection threshold than can be attributed to eye integration effects alone.
The effects of amplitude quantization on contrast threshold over a range of noise
conditions revealed that quantization levels below the observer's contrast threshold
had no effects on performance. An evaluation of azimuth data compression techniques
indicated that a predictable threshold dependence was found, based on the mean and
standard deviation of the background statistics. An investigation of the operator's
ability to detect a uniform target of fixed but unknown dimensions revealed a
contrast threshold dependence proportional to the square root of the target area,
with contrast threshold sensitivity decreasing for large target areas.
This paper presents the results of an experiment that studied human
recognition of infrared images. An experiment is described in which human
observers were asked to to discriminate between different types of modern armored
vehicles at various resolutions. In the original study', Johnson was concerned
with the four criteria of detection, orientation, recognition, and
identification, and a limited number of objects were used. This experiment used
many more vehicles than Johnson, but concerns only the tasks of identification
friend or foe (1FF), and identification. The vehicles are ones that would be
commonly encountered in a modern day confrontation between NATO and Warsaw Pact
forces. Simulated infrared images of these vehicles were presented to trained
observers and the resolution threshold for the identification friend or foe task
The solar spectral response of a Schottky-barrier infrared detector (PtSi) is compared to a photovoltaic detector (InSb) of comparable average over the region of interest. The effect of sunlight on a Schottky-barrier detector is found to be even more pronounced than that of a photovoltaic detector of equivalent sensitivity. To reduce this exaggerated effect, a spectral model has been developed for the selection of the optimum filter for use on a PtSi focal-plane array. The model makes it possible to reduce the effect of solar irradiance while not significantly impacting the performance of the detector against a thermal target. The spectral ratio of solar irradiance to a 290-K target for a Schottky-barrier detector is discussed. The solar-to-target contrast ratio is plotted against the integrated 3-5-micron spectral sensitivity to help the user select an optimum cut-off wavelength.
Imaging systems operating in the thermal infrared bands (3-to 5-urn or 8-
to 14-urn) are key elements in major electro-optical weapons systems.
Imager response as a function of radiative difference betwen target and
background is commonly expressed in terms of temperature difference, or
thermal contrast, between the target and the background. This can be
done since radiative difference and thermal contrast, under the
assumption of identical target and background emissivities and nearambient
temperatures, are linearly proportional. Thus FLIR detector
response is commonly expressed in terms of "minimum detectable
temperature difference" (MDT) and "minimum resolvable temperature
difference" (MRT) . Models such as the CCNVEO Static Performance Model
for thermal imaging systems, which uses thermal contrast as a target
characteristic, have had mixed success in predicting FLIR performance in
field tests. Experimentally measured smoke/obscurant transmittance
thresholds required for obscuring targets from FLIRs have large standard
deviations but tend to agree with the the Static Performance model. The
differences between model predictions and experimental results generally
have been ignored because a large number of variables in the tests (such
as variation in human response) cannot be controlled. Smoke/obscurant
countermeasures tests and calibrations of targets used in these tests
have been based on the assumptions that the emissivities of targets and
backgrounds were identical, constant with wavelength, and near unity.
However, this paper shows that relatively small differences between
target and background emissivities can lead to significant differences
between thermal contrasts predicted using true target and background
radiative differences and that predicted by brightness temperature
difference. Thermal contrast incorrectly estimated using an emissivity
of 1 for target and background (brightness temperature) and expected
thermodynamic temperatures can lead to major errors in the Static
Performance model's estimate of the transmittance level required to
reduce the detection/recognition capability of target observers using
FLIRs. This source of error may help explain the wide variation in
existing smoke/obscurant threshold data for FLIR5. The purpose of this
paper is to evaluate the effect of differences in target and background
ernissivities on thermal contrast estimates of radiative difference
between target and background, and to examine the effects such errors
have in estimating the transmittance threshold required to obscure a
target viewed with an imager.
The effects of pixel area and the total number of pixels in a staring Focal Plane Array (FPA) is mathematically treated to determine the optimal relationship between these two parameters. This presentation is a report of ongoing activities in this area. Current models for these devices are, in general, one dimensional treating the horizontal and vertical components separately. This analysis will attempt to define the Minimum Resolvable Contrast (MRC) performance criterion in terms of a two dimensional model.
The Multiband Electro-Optical Scanner is an experimental laboratory demonstrator, the purpose of which is to investigate the feasibility of missile detection using a scanning thermal imager in a number of discrete wavebands. The demonstrator utilizes three mid-IR wavebands and one FIR waveband to discriminate spectrally between real targets and false background clutter. The narrow instantaneous fields of view of these four detectors are scanned using broadband reflective optics common to all four beams, to provide a field of 3.5 x 10 deg. The signals from two reference blackbody sources are injected once per frame to provide radiometric calibration. The system design is described, illustrating the impact of optomechanical design on systems considerations.
The modeling and analysis of infrared target and background signatures continues to be a topic of interest
in the DoD. The question in many individuals minds is: What is the purpose of these signature prediction
activities? After all, the sensor perfonnance modeling community tends toward the use of simplistic target and
background representations in their models. Typically signature inputs to sensor models are nothing more than
target delta temperatures which assume homogeneous targets and backgrounds. This paper attempts to answer
the following two questions: 1 .Why do current sensor models have moderate to high fidelity modeling of the
sensor subsystems and systems performance and low fidelity input target/background signatures? and 2. Is a
high fidelity target and backgmund signature required to provide a meaningful estimate of sensor perfonnance?
Examples of current sensor models and predictive target and background signature models are provided and
discussed. A challenge is issued to the sensor modelers to learn more about signature prediction models and to
be innovative in their use in development of future sensor perfonnance methodologies.
Signal processing in the element (SPRITE) detectors are analyzed in terms of their main spatial-frequency-dependent parameters of modulation transfer function (MTF) and a number of equivalent elements. It is found that MTF depends on the length of the SPRITE element. The enhancement of signal-to-noise ratio also depends upon the element length and can be expressed in terms of a number of equivalent elements.
Since 1975, the NVL Static Performance Model for thermal Viewing Systems has been the primary tool in government and industry for evaluating thermal imaging system performance. However, advances in thermal imager technology have resulted in FLIR design improvements not specifically included in the NVL Thermal Model. Model development efforts at the U.S. Army Center for Night Vision and Electro-Optics have produced a model that includes features typically found in advanced FLIR systems, in particular sampling and improved vertical resolution. The Advanced FLIR Systems Performance Model (AFSPM) is a documented and validated computer model that predicts noise equivalent temperature, modulation transfer function, and minimum resolvable temperature for many thermal imaging systems.
As the price of testing infrared imaging systems in the field
rises and the price of fast desk top computers falls, imaging system
simulation becomes an attractive alternative to field testing.
C2NVEO's model development program uses human observers in
perception tests to gain insight to the effects of various FLIR
designs on the person's ability to do a task. The person must view a
variety of thermal targets through a variety of FLIRs. To meet the
perception test's need for many target/FLIR situations a FLIR
simulation computer program is being developed in parallel to the
model development effort.
The FLIR simulation is a spatial domain math based image
manipulator which uses high quality radiometric images and FLIR design
parameters as inputs. The program degrades the input image according
to the FLIR specifications to produce the output image.
This report describes the FLIR simulation computer program at its
current level of development, and presents some of the results.
This paper presents the results of initial testing of a computer simulation model
which has been designed for use by persons not necessarily expert in all or any of the
disciplines involved in signature analysis, and which can be utilized for evaluating sensor
system performance and prediction of sensor acquisition ranges. The ATIMS III airborne
turret infrared measurement system, ifight expenments, and the IASPM simulation model
are described. Analysis of preliminary results comparing experimental data with
simulated data for the 2 - 5 and 8 - 12 micron IR bands reveal the potential of the model
for simulating a multitude of sensor-observed phenomena. Model strengths and
shortcomings are discussed.
The prcblem of a1iasir is catn to all video imagirg systems,
both IR and visible. The 30 Hz frame rate (P5-170) places an upper lirait on
frequency, so that repetitive enoiena that vary nore rapidly than 30 Hz
will be aliased. For scenes where irvtion is involv1, the presence of
aliasir is often thvious, as the ige will be distortel. Harever, when
only teTrerature is varyirxj, the effects of aliasirx are difficult th
detect. In fact, usir a conventional IR iitager, one cannot diStirUiSh
whether a target is at thermal uilThrium, or its surface temperature is
varTing faster than the fraite rate.
We have develop1 a system which allcs repetitive thennal eventh at
frequencies as high as 2 kllz to be flrg1 without aliasirx. The system
functions as a video lock-in amplifier in that temperature caiponents of the
scene which vary at a preselected frequency are anplifi&1, while other
frequency ccaiponents of the target temperature average out over tine. A
prototype unit, consisting of an unndified scanning imaging radiareter
(Inframetrics IR-600), a E-T car!patible microcatputer, axxl specially
developed interface hardware arxl software, has been built ard tested.
A fast recursive convolution for the precise digital simulation of imperfect lenses and motion blurring of scanning detectors was invented for the production of large sets of human and Automatic Target Recognizer (ATR) test imagery. The method shows a wide range of other possible applications. A robust algorithm for image segmentation, the explanation of horizontal-vertical preference and high periodicity acuity in the human visual system might emerge.
Forward-Looking InfraRed (FUR) and TV sensors, laser rangefinders and beamriders have now
reached enough maturity to be integrated in Electro-Optical (E-O) systems used in air systems such as the Air
Defense Anti-Thnk System (ADATS ) produced by Oerlikon Aerospace Inc and selected by the U.S. and
Canadian Armed Forces as their low-level air defense system. When testing such systems, flexibility,
reproducibility and automation must be considered, especially when performing complex tests such as
boresighting, coded laser beam mapping, target tracking, laser ranging, sensor resolution and so forth. We have
designed and developed an optical test bench, the E-O Test Station* (EOTS), to perform these tests in the factoiy
during system integration and acceptance.
The EOTS consists of a special one-meter aperture f/3 collimator equipped with an advanced Focal
Plane Test Unit (FPTU). The station can be moved in the factory from one test bay to another, be aligned with a
Unit Under Test (UUT) and automatically perform various E-O production tests in a full laser safety and
controlled environment. In order for the EOTS collimator to meet the microradian alignment accuracy
requirements, we have designed and developed a special AutoCollimator System (ACS) used in the Company
Quality Assurance Department as the primary EOTS calibration standard. We also designed and developed a
Laser Alignment Periscope System (LAPS) installed inside the EOTS and used by the operator to ensure that the
collimator alignment is maintained during testing.
The EOTS and the ACS have other applications as well, like testing retrorefiection effects of
integrated E-O systems used in tanks, helicopters, etc., and in civilian applications such as the calibration of laser
intersatellite communication systems and the testing of lidars used in pollution monitoring. A variant field
version called Sensor Test Station is under development. This project was partly funded by the Canadian
With the advent of modern optical design programs featuring statistical tolerancing capabilities, the task of tolerancing multi-element optical systems has been simplified. However, the requirement for innovation still remains to ensure that the impact of all relevant environmental effects, not normally accounted for by the computer design program, is established. In this paper, a tolerancing methodology for an IR optical telescope is presented which combines the computer generated optical tolerances with the anticipated environmental perturbations by employing sensor defocus as the principle optical figure of merit. Included with the analysis is the apportionment of the optical tolerance budget to the individual opto-mechanical components and a discussion of the optical test and metrology program employed to systematically verify the as built telescope performance. Finally, a comparison is presented of the predicted versus as built optical performance.
A method has been developed, whereby the optimum focus of an IR telescope may be determined utilizing self imaging. A set of gold-bars, located adjacent to an array of detectors within a cooled focal plane, are reimaged and scanned across the detector array. The respective spatial and temporal frequencies of the gold bars and external scan mirror are selected to maximize the response of the sensor electronics. The resulting response to this cooled periodic stimulus is Fourier transformed and stored in memory for each location of the focal plane. The optimum focus is determined to be the position where maximum power at the stimulating frequency was obtained.
Thermal Imaging Systems are characterized by various tests such as Minimum Resolvable Temperature (MRT), System Intensity Transfer Function (SITF) and Noise Equivalent Temperature Differential (NETD). Numerous sources of errors can effect these test giving misleading results. These error sources are analyzed and a new correction methodology is presented.
Methods for focusing thermal imaging systems are discussed including visual inspection of resolution targets and edges, maximizing the system response to slits or periodic targets, maximizing the response at a particular spatial frequency, and maximizing the output of an edge detection algorithm. Visual inspection of a sweep frequency is found to be the most efficient method for determining focus in raster scanned systems.
This paper will present the results obtained at the Center
For Night Vision and Electro-Optics of laboratory measurements of
staring array thermal imaging systems. The results that will be
reported are a compilation of a number of different systems
tested. This data will be utilized not so much as a presentation
of system performance but to show the relevance and validity
of the tests. It is hoped that through the presentation of this
data the thermal imaging community will get a better understanding
of both laboratory and field performance of staring array
thermal imaging systems.
Real bodies emitting electromagnetic radiation in JR spectral re-
gions are imaged by infrared optical systems on detectors. An important point is to design these systems so that they work in a wide ternperature
range without refocussing. Athermic lenses fulfilling this
requirement are realized by specific correction measures. On this basis
a complete series of lenses for the short-focal-length range has
Two methods of making an objective measurement of MRTD have been under development. The techniques and some results obtained using different types of thermal imager, are described and discussed. Conclusions are that objective measurement techniques can now usefully replace subjective measurements as a production test.
A portable collimator and MRTD target system has been developed which provides a rapid go/no-go means of checking the performance of FLIR's installed on aircraft, tanks and other vehicles. The system is also suited to production testing.