The development process of new FPA imaging systems should be accompanied by an operational research study. The tool for such a study should be a model that predicts the performance of the overall system (detector, optics, signal processing, human observer), together with the target signature characteristics and the background properties. This model should yield a figure of merit that will be used for the performance study during the design and development process of the system. The influence of various parameters that will be used in the design process of the system can be studied using this tool. This work presents an approach where an image based sensor model was used to study the ability to detect and recognize different targets at various scenarios. The sensor model was used to simulate images of bar patterns for the evaluation of the modeled sensor MRTD. The results were compared to FLIR 92 predictions and the real sensor MRTD measurements. The model was then used to simulate targets embedded at various types of backgrounds. The images were presented to human observers that determined whether they detect or recognize the targets. This paper will bring a short description of the FPA sensor model and will present the methodology of using the simulation for sensor performance study. An analysis of some of the obtained results will be included as well.

For small unresolved or partially resolved objects the prediction of the performance of an IR sensor system is dependent not only on the optics and the atmosphere but also upon the geometry of the sensing elements within the detector. In particular this paper investigates the apparent scintillation caused by subpixel displacements of the detector from the optimum position is investigated. The effect is linked to the system point spread function and indeed the shape and size of the object. A simulator has been developed which can evaluate the effect of the shape and size of the sensing element, the optics, the viewed object and the atmospheric transmission, emission and background radiance. With the simulator it is possible to evaluate the effect that vibrations or pointing accuracy will have on the performance of the sensor systems for different types of targets. Results are presented which show that the variation in contrast of objects due to the shape and size of the active portion of a detector and the position of the target projection on the detector can be significant and should be taken into account when estimating system performance.

An extensive effort is ongoing to validate the TARDEC visual mode (TVM). This paper describes in detail some recent efforts to utilize the model for dual need commercial and military target acquisition applications. The recent completion of a visual perception laboratory within TARDEC is a useful tool to calibrate and validate human performance models for specific visual tasks. Some validation examples will be given for low contrast targets along with a description of the TVM and perception laboratory capabilities.

The ultimate goal of end-to-end system modeling is to simulate all known physical effects which determine the content of the data, before flying an instrument system. In practice we approach this ideal but do not attain it. In remote sensing, one begins with a scene, viewed either statically or dynamically, computes the radiance in each spectral band, renders the scene, transfers it through representative atmospheres to create the radiance field at an aperture, and integrates over sensor pixels. We have simulated a comprehensive sequence of realistic instrument hardware elements and the transfer of simulated data to an analysis system. This analysis package is the same as that intended for use on data collections from the real system. By comparing the analyzed image to the original scene, the net effect of nonideal system components can be understood. Iteration yields the optimum values of system parameters to achieve performance targets. We have used simulation to develop and test improved multispectral algorithms for : (1) the robust retrieval of water surface temperature, water vapor column,and other quantities; (2) the preservation of radiometric accuracy during atmospheric correction and pixel registration on the ground; and (3) exploitation of on- board multispectral measurements to assess the atmosphere between ground and aperture. We have evaluated the errors in these retrievals for a variety of target types due to: telescope OTF, calibration bias, system noise, spacecraft motion and jitter, atmospheric effects, telescope distortions, and co- registration during processing of multispectral images with offset pixels.

This paper describes the development of a computer based infrared mission rehearsal system (IR-MRS). The IR-MRS can be used for IR mission planning, selecting navigation waypoints, and for mission rehearsal. For specific scenarios, locales, and environments the IR-MRS can simulate ingress/egress corridors and realistic target engagement ranges. In addition, the sensor model's modular design allows the IR-MRS to have a very wide applicability to problems involving either the design or the simulation of infrared sensors. The simulation is performed on Silicon Graphics (SGI) computers in order to take advantage of SGI's accelerated 3D graphics hardware. The system capability includes, conversion of synthetic visual databases to infrared databases, environmental effects, IR database fly through, and modeling the characteristics of specific FLIRs.

Wavelet transforms are currently being used for a number of applications such as cue feature and noise extraction form images and acoustic signals. The objective of this paper is to describe and apply the author's algorithm that uses wavelets for finding the clutter in infrared and visual images. Once the clutter is found, the probability of detection is calculated. The Reynolds identity and Tidhar's and Rotman's probability of edge metric are extended to encompass the wavelet methodology for multiscale clutter metrics in IR and visual images.

One of the larger sources of variation in human performance predictions is observer-to-observer variability. Whether because of differences in experience, training, motivation, innate ability, or perceived risk, observer performance varies. This paper explores observer variability in the task of detecting military targets in a FLIR image. Data for the analysis were obtained from a recent perception experiment conducted by NVESD utilizing 36 observers examining thermal imagery. Using the aggregate performance of the group as a measure of target difficulty, this paper compared the individual observer performance to that of the group. It was obvious that individual observers did not have identical performance characteristics. FOr example, targets detected by 50 percent of the group were detected by 20 percent of the observers over 70 percent of the time while another 20 percent of the observers detected then less than 20 percent of the time. Thus, the distribution of observer performance is fairly broad. This analysis of observer variability has implications for the modeling of target acquisition in combat simulations such as CASTFOREM and JANUS. These simulations currently use a uniformly-distributed, randomly-assigned threshold approach.

This paper discusses the impact of display pixel shape, size and spacing on displayed image quality. The application of FOurier transforms to sampled imagery is described. A characterization function for sampled imagers is derived by examining the image formation of a point source. Misapplication of the sampling theorem is also discussed.

The Aerospace COrporation has developed a general purpose simulation tool for modeling visible and infrared surveillance and imaging systems. VISTAS (visible and infrared sensor trades, analysis and simulations) has been used to model both scanning and staring systems, and to compare their performance against identical scenarios. The simulations begin with a high resolution background scene, which may be either real data or a synthetic construct. If desired, targets may be embedded in the scene. The next steps in the simulation include degrading and reformatting the scene according to the specifics of the sensor system transfer function, which includes the optical point-spread-function, detector aperture, temporal aperture and filter response. The process continues with the addition of noise and the scene is resampled to the sensor resolution, including the effects of multiple samples per dwell and cross-scan oversampling. FOr time-dependent simulations, jitter and platform drift are included to produce a sequence of output images which represent the sensor response. In this paper we will show how the modeling of systems can be used to provide feedback to the design stage through tradeoff analyses. For example, scanning systems can improve their performance by oversampling, but pay a penalty in higher data rates and potentially noisier output if the integration time must be lowered. Staring systems that utilize a frame different algorithm to detect moving targets can benefit by reducing jitter, but the increased costs associated with providing low jitter must be weighted against the potential benefits. Showing the system designer the results of tradeoff analyses has proven to be a substantial aid in the design of a sensor for a given mission.

Post correction uniformity (PCU) schemes are employed in some existing imaging systems to correct spatial nonuniformities. The focal plane arrays in the imaging system may either be a scanning or a staring focal plane array. PCU specifications may be included as part of the full FPA specification. Current PCU specs typically designate a level of fixed pattern noise relative to temporal noise at various selected background temperatures. The selected temperatures are within the expected range of target temperatures to be seen. An issue seems to be the difficulty of keeping the post correction fixed pattern noise levels within specified levels at all specified background temperatures. This paper examines the effectiveness of the tow point linear gain-offset correction method for systems with nearly linear response, and relates post correction uniformity to MRTD performance through the 3D noise model. Such a relationship is expected to facilitate FPA developers in specifying FPA PCU requirements, tracking and identifying sources of MRTD failures in FPAs, and quantifying the benefits of this particular PCU scheme.

An analysis methodology and corresponding analytical tools for rapid top-down design of multi-spectral imaging systems is presented. Beginning with top- level customer-dictated system performance requirements and constraints, the critical system and component parameters in the electro-optical image chain are derived, performance analyzed, and iterated until a preliminary design that meets customer requirements is generated. System parameters and components composing the image chain for staring, scanning, pushbroom, and time-delay and integrate systems include: aperture, focal length, field of view, cold shield requirements, image plane dimensions, pixel dimensions, pixel pitch and fill factor, detection quantum efficiency, optical filter requirements, image plane dimensions, pixel dimensions, pixel pitch and fill factor, detection quantum efficiency, optical filter requirements, and temporal sampling parameters. The performance analysis is accomplished by calculating the imaging system's optical response (to a scene radiance), total noise, and imaging resolution. The noise components include photon noise due to signal, scene and atmospheric background, cold shield, out-of-band optical filter leakage and electronic noise. System resolution is simulated through cascaded optical transfer functions (OTF's) and includes effects due to atmosphere, optics, image sampling, and system motion.

Models of target acquisition by human observers are probabilistic. Two reasons for the indeterminate nature of the models are (1) the observer ensemble includes individuals with varying levels of competence in target acquisition, and (2) the target acquisition process is itself (at least in part) a random process. Each of the two commonly used combat simulations, CASTFOREM and JANUS, assumes that (a different) one of these completely accounts for the indeterminacy, and ignores the other. While the distinction is irrelevant for the simple one-on-one case, it has recently been shown that the choice can affect the outcomes of more realistic many-on-many engagements. We propose a model in which target detection probability is a function of both a target statistic and an observer statistic. Our analysis of recently compiled observer test data validates the model and provides it with the correct quantitative balance between variations among observers and the inherently stochastic component of target acquisition.

Both edged-based and textural-based metrics can help us understand the effect of clutter in images that a human observer is searching in order to acquire targets. By studying the actual fixation points of the observers, we can determine which features distract or confuse an observer in his attempt to acquire the true target.

A recently developed model of a spin-scanning infrared seeker increases the accuracy of signal processing evaluations and seeker performance predictions compared with previous models. The simulation includes the effects of targets and backgrounds, seeker scanning and detector signals, gyroscope dynamics, signal processing, and acquisition and tracking. The model also incorporates realistic representations of noise, look-angle-dependent optical blur, gyroscope errors, and scan modulation. This modulation is caused by magnetic coupling from the gyroscope system and by radiation from the seeker dome and internal components. The result is artifacts on the detector waveforms that interfere with target detection. Aerothermal and laser-heating tests have been conducted to measure the optical scan modulation from the dome and internal components. Seeker waveforms as well as dome and component temperatures were measured to provide data for modeling the heating effects. These measurements are used to predict performance degradation caused by aerothermal heating of the seeker in flight.

This paper describes the results of experiments that were conducted in order to characterize the types of noise limiting the performance of an amber InSb charge injection device focal plane array (3-5 microns) of 256 by 256 pixels. This is part of the work done at the Defense Research Establishment Valcartier to develop a wide-area-coverage infrared surveillance system. The emphasis is put on the analysis of the postcorrection spatial noise that reduces the array sensitivity to weak point-source targets. This residual noise limits the improvement provided by an increased array integration time. Furthermore, the results show that a temporal low frequency noise component has a more severe effect than detector nonlinearities. However, this problem can be partly resolved with a periodic offset compensation obtained by reference image subtraction. The reference image is acquired when the blade of a flat black chopper wheel completely blocks the aperture of the camera. The chopper wheel is synchronized on the acquisition process. Results show that this compensation method can efficiently reduce the low frequency noise level and enhance point-source target detection.

A field trial has been conducted to obtain range performance data for two staring IR imagers and for a scanning first generation IR imager. Qualitative observer trials were made in the field. Perception experiments with the recorded imagery were made after the trial to obtain statistically relevant data. The field trial and the perception experiments are presented together with the observer results. Using the range performance of the scanning imager as a reference, the results obtained with the staring systems are compared with predictions made by range performance models. It is shown that a range model based on an MRTD which is limited to half the sampling frequency (Nyquist frequency) does not agree with the experimental results. Good predictions are obtained, however, with a range model based on the MTDP (minimum temperature difference perceived).

Tactical air-to-air or ground-to-air IR missile seeker design involves a vast number of interrelated design criteria spread over a very large parameter space. To further exacerbate matters, many of these critical parameters are dynamic in nature. A common approach to seeker design is to model the behavior of each critical component, or set of components, with a mathematical simulation of its predicted behavior. While this approach is generally very high in fidelity, it tends to be very computationally (and man-hour) intensive, requiring a large number of iterations for each component. Furthermore, many of these modeled components, for example, signal processing, are both space- and time-dependent in nature, and hence are dynamically related to the missile flyout (dynamics) model; consequently, they cannot be fully modeled by a purely static model. An end-to-end (launch to closest approach) dynamic seeker performance model is briefly described herein, which allows for rapid trade space realization over all relevant seeker parameters, and an example seeker design flowdown using the dynamic performance model is presented.

Members of the electro-optics community have essentially redefined the modulation transfer function (MTF) for focal plane arrays in different ways. Although these particular discrepancies make little difference below the Nyquist frequency, they are significant for system which push limits and process above-Nyquist frequency data. This issue is briefly reviewed, and it is shown that consistency with the Fourier method of defining MTF for focal plane arrays implied sample phase averaging, separation of the transferred sinusoid remnant with the input frequency from those of alias frequencies and recognition that the contrast ratio expression for MTF is a special case. Differences in the implications of discrete sampling analysis and the sample- and-hold operation are presented. In this approach MTF and alias effect are separated so that MTF represents a single-valued, global transfer entity and alias is defined by two different characterization metrics.

As staring focal plane array (FPA) detectors become more readily available, imaging IR sensors can be constructed in more compact packages that are lighter and consume less power than first or second generation scanning IR sensor packages. However, FPA detector-based imagers typically demonstrate reduced resolution when compared to scanning systems with similar instantaneous-field-of-view. This resolution limitation is created by a the active pixel size and the spatially synchronous scene sampling native to staring FPA systems. A technique called microscanning can be used to improve the resolution of staring systems by over-sampling the scene; moving the image of the scene on the detector in a controlled fashion. This paper presents a compact MWIR staring FPA airborne forward looking infrared sensor design using microscanning for resolution improvement.

Microscanning is a technique that allows to double the resolution of a given staring array imager. It consists in taking multiple images of the same scene while displacing each time the image over the detector plane by a distance equal to a fraction of the detector pitch. The technique is limited by the time required to shift the image from one point to the other and by the precision of the movements. This article describes work that was done under contract for the Defense Research Establishment Valcartier as part of the Wide Area Coverage Infrared Search System (WACISS) project to develop a fast microscanning imaging device. The system includes three main sections: the microscanning head, the controller and the power amplifier. THe microscanning head is made of a lens and a two-axis microtranslation table driven by two piezoelectric translators. The controller drives a high voltage power amplifier which in turn drives the translator. The controller allows four operation modes: fixed position, 2 X 2, 3 X 3, and 4 X 4 microscan. It works in open as well as in closed loop for precise displacements. The systems will be integrated to the WACISS project and will serve as an aid for the identification of detected objects.

Current infrared imaging systems used for surveillance and search and rescue operations possess two fields of view which may be alternately selected by the operator: a wide field of the order of 20 degrees is used for the search and detection of targets, and a narrower field of a few degrees is selected for the recognition tasks. However, the degraded sensitivity and resolution of the wider field prevents it from fulfilling its function adequately. A new concept based on the focal plane array detector technology is intended to correct this drawback and to improve future infrared surveillance system for search and rescue operations. Simulating the properties of the human eye, the concept allows the simultaneous surveillance and image acquisition in two fields of view. A wide peripheral field of view (60 degrees) with increased sensitivity but lower resolution is dedicated to search and detection. A narrower field (6 degrees), which can be steered within the wider field, allows the recognition of detected objects with an improved resolution obtained by the use of microscanning techniques. THe high resolution required for the simultaneous display of both fields of view has led to the development of a new type of display, based on optical projection and superposition, better adapted to the human eye and hence optimizing the human interface. The constraints on the opto-mechanical and electronic design imposed by the mobility of the narrower field within the larger one, the microscanning mechanism and the calibration requirements of the focal plane array are discussed, and the selected solutions are presented. The limitations of the system in its present state of development are exposed and the plans for future improvements are elaborated.

The ImagIR infrared imaging system along with image based test methods has been used as a primary approach for evaluating focal plane array performance at Santa Barbara Focalplane (SBF). This information has been used to provide feedback to the in-house processing line as part of the SBF continuous measurable improvement program. Recently, the testing capabilities at SBF have undergone significant expansion through an elegant integration of the ImagIR systems with other commercially available software and hardware. The resultant test set contains a new powerful test environment where system imagery and traditional radiometric figures of merit are readily available for evaluation and correlation. Key features of this environment include a user friendly interface, flexibility, and low development, replication, modification, and maintenance costs.

A method based on the line scan of a narrow slit pattern is described for MTF and MRTD determinations. A digital signal analyzer and automatic computer worksheet are utilized for executing this methodology. This single-scan procedure can replace a series of measurements with different sized bar patterns, and it facilitates rapid imaging system characterization for weapons scheduled for intensive hardware-in-the-loop testing. The challenge of this method was to overcome the practical data processing problems encountered. MRTD determinations using the slit-scan method agree very well with conventional bar pattern measurements and has been validated for a variety of imaging sensors. The automatic worksheet analysis determines MTF and both the objective, line scan MRTD as well as the 'eye-brain MRTD' based on a human vision model.

A radiometric calibration station (RCS) is being assembled at the Los Alamos National Laboratory (LANL) which will allow for calibration of sensors with detector arrays having spectral capability from about 0.4-15 micrometers. The configuration of the LANL RCS is shown. Two blackbody sources have been designed to cover the spectral range from about 3-15 micrometers, operating at temperatures ranging from about 180-350 K within a vacuum environment. The sources are designed to present a uniform spectral radiance over a large area to the sensor unit under test. THe thermal uniformity requirement of the blackbody cavities has been one of the key factors of the design, requiring less than 50 mK variation over the entire blackbody surface to attain effective emissivity values of about 0.999. Once the two units are built and verified to the level of about 100 mK at LANL, they will be sent to the National Institute of Standards and Technology (NIST), where at least a factor of two improvements will be calibrated into the blackbody control system. The physical size of these assemblies will require modifications of the existing NIST Low Background Infrared (LBIR) Facility. LANL has constructed a bolt-on addition to the LBIR facility that will allow calibration of our large aperture sources. Methodology for attaining the two blackbody sources at calibration levels of performance equivalent to present state of the art will be explained in the paper.

Recent requests for test stations to characterize and evaluate thermal and visible imaging systems have shown remarkable similarities. They contain the usual request for target patterns for the measurement of MRTD, NETD, SiTF for the infrared thermal imager and similar patterns for measuring CTF and SNR for the visible imager. The combined systems almost invariably include some type of laser designator/rangefinder in the total package requiring the need for LOS registration among the various individual units. Similarities also exist in that the requests are for large collimator apertures and focal lengths for projecting the desired signals into the unit under test apertures. Diversified Optical Products, Inc. has developed and is continually improving test station hardware and software to provide modularity in design and versatility in operation while satisfying individual test requirements and maintaining low cost. A high emissivity, DSP controlled, high slew rate, low cost, blackbody source with excellent uniformity and stability has been produced to function as the driver for thermal image target projectors. Several types of sources for producing energy in the visible portion of the spectrum have been evaluated. Software for selection of targets, sources, focus and auto- collimation has been developed and tested.

The JTAMS program has developed calibration and data analysis procedures to improve the data quality from tactical missile signature tests. These procedures include a combination of laboratory and field test calibrations as well as quick-look analysis procedures which are used to identify inconsistent or erratic instruments during the test execution. The procedures have led to significantly higher data quality as evidenced by signature measurements carried out in the course of the three-year program.

Determination of the modulation transfer function (MTF) of an imaging system can be tedious and laborious when done by conventional means. However, through clever manipulation of test imagery, the line spread function and hence the MTF can be more easily found. This paper will introduce a new data reduction approach which is a viable candidate as a new metric for testing imaging systems in the field. The repeatability and objectivity of this analysis appears to have great potential utility.

Often FLIR system performances are predicted using standard models such as FLIR92 and ACQUIRE from the US Army NIght Vision and ELectronic Sensor DIrectorate. These models typically produce results such as MRTD verses spatial frequency, NETD, and probability of detection and/or recognition ranges, etc. This paper presents and end-to-end FLIR system performance model using actual IR scene as an input, and produces visual output. This model is developed and run using a commercially available PC based scientific spreadsheet, and it can predict how an actual target would appear on the display of a particular FLIR sensor under a given atmospheric condition and range. It serves as a helpful supplement to the FLIR92 and ACQUIRE models.

Engineering perspectives related to specification, design, manufacture and evaluation of collimator lenses are discussed using the example of a recently manufactured collimator for use in thermal imager evaluations. Collimator optics are important test facility components used for the evaluation of thermal imager systems. The thermal imager test facility at NVESD needed new collimator lenses in order to use targets small enough to work within the limitations of blackbody uniformity while testing thermal imager systems for the Driver's Viewer Enhancer program. These systems require very low spatial frequency targets to completely characterize minimum resolvable temperature difference. Three perspectives on the design, manufacture and testing of a collimator lens assembly are presented in this paper: (1) the perspective of the user, who is the optics specifier; (2) that of the optics designer/manufacturer; and (3) that of the optical testing engineers who validated the optics design using CODE V and tested the optics at NVESD's optical test facility. Different perspective result in the customary use of different parameters to define specifications. Therefore, optics specifiers must consider various perspectives to ensure the lens design meets desired requirements.

There are a large number of EO and IR sensors used on military platforms including ground vehicle, low altitude air vehicle, high altitude air vehicle, and satellite systems. Ground vehicle and low altitude air vehicle (rotary and fixed wing aircraft) sensors typically use the probabilities of discrimination (detection, recognition, and identification) as design requirements and system performance indicators. High altitude air vehicles and satellite sensors have traditionally used the National Imagery Interpretation Rating Systems (NIIRS) performance measures for guidance in design and/or measures of systems performance. Data from the high altitude air vehicle and satellite sensors is now being made available to the warfighter for many applications including surveillance and targeting. National imagery offices are being merged and restructured to more fully support warfighters and connectivities to high altitude air vehicle sensors. It is becoming more apparent that the gap between the NIIRS approach and the probabilities of discrimination approach will have to be addressed. Users, engineers, and analysts need to have a comparative basis for assessing the image quality between the two classes of sensors. This paper describes and compares the two approaches.

Standardization of intermediate test support equipment has been successful in reducing overall support cost by minimizing manpower and eliminating redundant capabilities. This paper addresses future improvements that further support the goal of reducing life cycle cost. Future test needs are presented based on emerging technologies in electro-optics which provide linger range target detection and identification. Meeting these test needs is critical in maintaining the greater levels of capability for the future.