A Transportable Direct Write Scene Generation (TDWSG) test capability has been developed at the Arnold Engineering Development Center (AEDC) for visible and IR focal plane array (FPA) testing which utilizes laser sources and two-axis acousto-optic deflectors. The objective of this effort is to provide a test and evaluation facility which will help reduce space sensor development risks by testing FPAs with their data subsystems against realistic mission scenarios in a space environment. The TDWSG's performance envelope covers both high- speed (100 microsecond(s) frame time) scanning and slower staring formats. A modular concept is used to address large (512 X 512 pixel) FPAs. Scene inputs can be derived from various sources including the Strategic Scene Generation Model (SSGM). A continuance of this effort is being applied toward development of a fixed-site Scene Generation Test Capability (SGTC).

This paper will provide an overview of the critical Infrared (IR) projector requirements and specifications for Hardware-in-the-Loop (HWIL) simulations of imaging IR systems and will review the most prominent technologies associated with IR scene projection. Each method will be briefly discussed in terms of physical operation with all relevant advantages and disadvantages being highlighted. A comparison will be provided of the different categories of projection devices for applicability to HWIL simulation.

A Scophony Infrared Scene Projector (IRSP) was developed for use in evaluating thermal- imaging guidance systems. The IRSP is a very high frame rate, laser scanned projection system incorporating Scophony modulation. The Scophony IRSP serves as the image projection system in the Kinetic Kill Vehicle Hardware in the Loop Simulator (KHILS) terminal guidance simulation. It is capable of projecting multiband target engagement scenarios with high fidelity using Aura's proprietary software/electronic control system. The Scophony IRSP utilizes acousto-optical (AO) devices to produce the required imagery at separate wavelengths, simultaneously. The separate scenes are combined and projected into the imaging guidance system. The Scophony IRSP has been installed and integrated into the KHILS facility at Eglin Air Force Base, Florida. Some performance characteristics of the IRSP have been measured. The current presentation provides a brief description of the Scophony IRSP and a performance evaluation. The performance characteristics measured are spot size, dynamic range, and field of view. Further characteristics may be reported as they become available.

The advent of missile seekers with dual-spectrum (simultaneous radar frequency/millimeter wave (RF/MMW) and optical) tracking systems has led to a requirement to develop the simulation tools necessary to test these systems. One of the most important tools for testing missile hardware is a full-seeker hardware-in-the-loop (HWIL) simulation. There is a substantial effort to improve the ability to generate and project RF/MMW imagery and optical imagery separately to systems under test. However, the dynamic real-time generation and simultaneous projection of realistic dual-band imagery to these seekers continues to be a major technological challenge. The purpose of this paper is to discuss methods and issues inherent in dual-spectrum seeker HWIL simulation.

A multi-technology high performance computing system based on the Open Parallel Architecture Design Specification (OPADS) platform is being evaluated for use as a graphics engine for spherical scene projection. This system is designed to make available the massive quantities of real-time processing power needed to support complete real time scene generation and projection of complex dynamical maneuvers for applications such as scientific visualization and three dimensional database creation and interaction. A comparison is also provided between head mounted projection systems and walk-in spherical scene projection systems.

The optical aspects of infrared projectors designed specifically for application within real-time hardware-in-the-loop simulation systems are examined. Particular attention is paid to the form of the spatial frequency transfer function and thereby to the factors that affect the spatial resolution within the composite optical system defined by the infrared projector and the imaging unit under test. Expressions are also developed allowing the calculation of both the output signal radiance and effective blackbody temperature for reflective infrared projectors, thus enabling the direct comparison of the radiant output capabilities of projectors of the emissive and reflective types.

We report the operation of a video-based infrared scene projector which utilizes an electron beam-addressed membrane light modulator, or e-MLM. The projector may be described as a schlieren projector which modulates both the phase and intensity of a readout light beam. In the present work, the readout source was a He-Ne laser and the readout wavelength was 3.39 micrometers . 128 X 128 pixel images were projected at a 30 Hz frame rate, and imaged by an infrared camera. The electrical input to the projector was a standard RS-170 signal provided by a commercial video cassette player. Some of the characteristics of the projector include flickerless framing, 100:1 contrast ratio, and continuous-tone gray scale. This technology is capable of high simulated temperatures and radiance. With further development, the technology should be capable of full NTSC resolution and higher frame rates.

Thermal background models are of particular interest in many fields of science and technology, e.g., meteorology, hydrology, agricultural management, remote sensing, and military applications. The complexity of the different modeling strategies varies to a considerable degree from very simplistic approaches for real-time applications to very complex research grade models. This paper summarizes the important physical and biological processes and introduces different modeling concepts. It pinpoints areas of lacking knowledge and shows possible solutions to the problems.

This paper deals with a computer program to calculate radiance map. A scene is described with a set of polygons. Each polygon has its own spectral reflectivity and its own temperature. Radiances and transmissions are calculated with LOWTRAN7. We explain our radiometric equation choice and how we use LOWTRAN7 code to minimize error calculation on transmission. We show comparison between photographs and simulated images for two different spectral bands (8 - 12 micrometers versus 3 - 5 micrometers ).

The diversity and complexity of background materials and clutter features in the infrared makes modeling backgrounds a more difficult effort than modeling target signatures. It may be appropriate to implant modeled target images into measured background images for simulation purposes because the target signatures are more reliably predicted and can be altered to match the measured background environment conditions. The extrapolation of measured target signatures to environmental conditions similar to those of a measured background image was described earlier. The extrapolated synthetic target image can be implanted into a real background image with radiometric consistency. Target implantation introduces the possibility of a number of non-physical artifacts, one of which is a difference in the resolution of the target and background parts of the image. Faceted target models contain highly resolved edges and internal features, while real background images contain blurring due to the measurement system. An approach is presented which blurs the internal detail of the target to match the known resolution properties of the background measuring instrumentation, and then adjusts the target-background boundary without introducing additional blurring to the background zone around the target. The resultant synthetic image is optically and radiometrically consistent with an image of a real target in the same background. Figures show details of target/background interfaces after resolution modification. Other artifacts illustrated include partial target obscuration by background objects and simulated motion of the target within the image.

The thin film resistor array choice of infrared projector technology is characterized by the comparatively large values of pixel fill factor and emissivity that can be attained but is limited by materials and heat transfer constraints. In this paper, the characteristics and limitations of an infrared projector test device based on a 2 X 25 bilinear thin film nichrome resistor array are described. The steady state and transient performance characteristics have each been assessed by use of both analytical and finite element heat transfer techniques. Test devices based on the design that gave the best predicted performance have been fabricated on a silicon wafer substrate by application of the conventional techniques of photolithography, etching and vacuum deposition, each array being comprised of a patterned polyimide insulation layer sandwiched between the substrate and the nichrome heating elements. Initial characterization experiments have demonstrated a 200 degree(s)C operating temperature capability and 10 - 90% rise and fall times of the order of 100 - 200 microsecond(s) . It is shown that the risetime can be improved significantly by application of a tailored drive voltage waveform, as can the falltime of appropriate thermal connection of the substrate to a low temperature heat sink. Modes of device failure are also discussed.

DOD has a requirement to develop and evaluate systems to detect targets against clutter backgrounds. This requires a comprehensive understanding of both target and clutter properties, such as spectral signatures, radiance distributions, edge effects, etc. Acquiring a complete database for all possible physical conditions and instrument parameters is not feasible. To supplement the measured database, Ontar, under two Phase I Small Business Innovative Research (SBIR) contracts to the Naval Surface Warfare Center, and the Naval Weapons Center has developed a workstation for generating radiance maps of complicated objects and performing integrated, multispectral, multitemporal image simulation. The system includes the capability to model targets, atmosphere, clouds, sea surface, natural and man- made terrain features. Currently, the package uses an aircraft model and a first-principle's cloud model. The software, written in FORTRAN, has a user-friendly interface and graphics capabilities. The current software operates on high end PC (80386 and 80486 machines). Results of integrated scene simulation and model validation will be presented.

The Strategic Defense Initiative (SDI) must simulate the detection, acquisition, discrimination and tracking of anticipated targets and predict the effect of natural and man-made background phenomena on optical sensor systems designed to perform these tasks. NRL is developing such a capability using a computerized methodology to provide modeled data in the form of digital realizations of complex, dynamic scenes. The Strategic Scene Generation Model (SSGM) is designed to integrate state-of-science knowledge, data bases and computerized phenomenology models to simulate strategic engagement scenarios and to support the design, development and test of advanced surveillance systems. Multi-phenomenology scenes are produced from validated codes--thereby serving as a standard against which different SDI concepts and designs can be tested. This paper describes the SSGM design architecture, the software modules and databases which are used to create scene elements, the synthesis of deterministic and/or stochastic structured scene elements into composite scenes, the software system to manage the various databases and digital image libraries, and verification and validation by comparison with measured data. The focus will be on the functionality and development schedule of the Baseline Model (SSGMB) which is currently being implemented.

Tactical Decision Aid codes provide field prediction of maximum range of FLIR use using simplified local environmental parameter input. A series of experimental comparisons at sea using airborne operational FLIRs with an instrumented ship target have shown poor correlation of observed range with prediction for detection and recognition. Classification and recognition range in UFLR are found to be highly insensitive to radiosonde atmospheric profile data input. Previous work has addressed modeling of the average target to background contrast temperature difference and atmospheric propagation of contrast. This paper addresses the implementation of the MDTD and MRTD algorithms in the code. Comparisons are presented of the prediction accuracy of the UFLR TDA using the standard Moser/Hepfer algorithm and an adaptation of the Johnson criterion used in the NVEOL Ratches code. For the limited data set of the study a reduction of RMS prediction error is achieved using the NVEOL algorithm.

Analytic expressions for the variance of the sea surface slopes used in the Cox-Munk surface models are derived for two wave spectra models currently used in IR sea surface models, i.e. the Pierson-Moskowitz and the Pierson-Neumann spectra. These relationships provide a basic coupling between the oceanographic community efforts in sea surface characterization and the IR modeling community that is developing sea surface radiance models such as Levesque and St. Germain (SPIE 1311), and the SEABEAM model of Ball (SPIE 1311).

The possibility of improving the image of a coherently reflected object under active illumination by a collimated laser beam through turbulence is experimentally investigated. It is demonstrated that when light travels to and from the object through the same inhomogeneities of the turbulent medium, the time-averaged image has higher quality than with independent propagation.

Methods of error analysis are described which were used to compare SPIRITS (Spectral Inband Radiance of Targets and Scenes) simulations with actual field test data of jet aircraft. Two distinct methods are discussed for spectral error comparisons. A simple quantitative method of comparing errors between simulated and actual IR images is also described. In addition, the requirements imposed on field test measurements and corresponding simulations to allow accurate comparisons are described.

A perturbation-theoretic approximation of the radiative transfer equation which neglects photon dispersion is used as a modelling basis for the propagation of the image of a self- luminous target through a turbulent atmosphere which also possesses inhomogeneously distributed turbidity along the propagation path. A contrast ratio is then introduced which provides an indictor of the relative contribution of the unscattered or coherent image component to that of the scattered or incoherent image component. Analytical expressions are then derived for the contrast ratio from the approximate form of the radiative transfer equation in the case of an inhomogeneously dispersed Joss thunderstorm rain distribution in the presence of turbulence. The case is clearly demonstrated for the need to consider a measure of the points of demarcation at which the dominant roles of the scattering processes due to turbidity and turbulence are exchanged. Such a measure can provide a performance parameter for the application of adaptive optics methods that are specific to the particular dominant scattering mechanism given the prevailing target size, total propagation length and overall propagation parameters.

Meteorological modules were developed to describe characteristic maritime scenarios in various oceanic areas for DREV complimentarity studies of shipboard defense. The best means of depicting the maritime atmospheric environment was found to be on the basis of air mass analysis. A methodology was developed whereby, through a mixture of man-machine objective analysis of upper air radiosonde measurements at the 850, 700, and 500 mb levels, typical airmasses could be identified. Characteristic scenarios were then defined based on physical considerations of air mass theory. Utilizing an extensive 10-year set of worldwide radiosonde, ozondesonde, and surface observations collected from a combination of land-based stations, oceanographic buoys, and weather ships, frequency and correlation statistics of various global and derived meteorological and oceanographic parameters were established for the CANLANT, NORLANT, WESTLANT, EASTLANT, IBERLANT, MARPAC regions, the ARCTIC OCEAN to 85 degree(s)N, the BALTIC SEA, MEDITERRANEAN SEA, PERSIAN GULF, RED SEA, GULF OF OMAN, and the INDIAN OCEAN. These descriptions included atmospheric profiles of pressure, temperature, dewpoint and relative humidity, wind speeds and direction, refractivity index, and ozone concentration from the surface to approximately 20 km., as well as associated surface visibility, clouds and weather, sea state, and duct height conditions. Many of the derived parameters were found to be a strong function of the defining airmass scenarios. The spatial distribution of these scenarios was also determined.

A nonintrusive experimental method, utilizing two Quad Cell detectors and a cross beam correlation (CBC) technique, was applied to Teledyne Brown Engineering (TBE) to investigate the relationship between fluctuating optical properties and image distortion caused by high velocity turbulence. A laboratory based Dual Nozzle Aero-Optic Simulator (DNAOS) was used to produce a mixing/shear layer that simulated a flight level aero-optic environment. Two Quad Cells were used to simultaneously measure the centroidal shift of two orthogonal laser beams that were located in a plane normal to the direction of flow and were coincident at only one point within the mixing/shear layer. The 2D angular deviations of the laser beams were calculated from the recorded centroidal fluctuations. From this data, the cross correlations could be calculated to determine the turbulence induced optical deviation (wavefront distortion) and density related index of refraction fluctuations for the flow field. The experimentally measured angular deviations were found to compare reasonably well to theoretically predicted values. This demonstrates that Quad Cell CBC can provide a nonintrusive method of accurately characterizing the fluctuating optical properties resulting from small scale hypersonic turbulent structures.

A multispectral passive sensor may be used to measure the temperature of an object provided that the light emitted is described by a black (gray) body distribution, and atmospheric attenuation coefficients between receiver and target are known. With additional time evolution passive measurements, it is possible to estimate target temperature even when attenuation coefficients are not known. In this paper we establish limits on the accuracy of temperature estimation using selected statistical models and LOWTRAN VII generated atmospheric scenarios. The results presented and the methodology developed are of fundamental importance in determination of the limits of infrared sensing. The framework presented has applications in multispectral scene analysis, thermographic mapping, and remote sensing.

The output from an infrared projector can be characterized in terms of either output signal radiance or effective blackbody temperature. In this paper, expressions from which both parameters may be derived are developed by spectral analysis of the infrared system comprised of an emissive infrared projector and an imaging unit under test (UUT). Features of the analysis include the separation of the infrared projector characteristics from those of the UUT, the calculation of the output signal radiance in terms of standard blackbody radiation formulae and the inherent algebraic connection that is maintained, enabling the UUT signal current to be readily determined once the output signal radiance is known. A dimensionless graphical procedure is developed for the subsequent determination of the effective blackbody temperature.

A medium power, narrow beam, electron gun is shown to produce a viable heat source for IR scenario simulation. The electron beam strikes a cryogenically cooled screen, producing a small infrared source against the cold background. Electronic deflection modulates and directs the beam, creating a real-time scenario simulation. Beam size and thermal spot spreading determine spatial resolution of the scene. Thermal relaxation time of the screen determines temporal resolution. Computer modelling is used to predict gun and screen performance. A series of experiments, used to confirm the predictions of simulation, and their results are described.

A computer model has been developed to determine the degradation of sensor performance due to contamination. These particulates may either by suspended in the near field-of-view or deposited on optical surfaces within the system. This system response simulation, PEARLSS (Particulate Effects Analysis on Response Levels of Spaceborne Sensors), is currently able to model many of the key dynamics of the particles to determine their distribution in the field-of- view and on the mirrors. Light from various sources is scattered off of the particulates. This radiation is mapped as a noise scene at the first mirror and the focal plane. The code is intended for performing parametric studies on design parameters such as sunshade geometry, system materials, temperatures, pre-launch cleanliness levels, and pointing directions. Sources of radiation currently modeled are the Sun and Earth, as well as user-defined sources. An overview of the simulation's approach to tracking contamination and scattered radiation is given.

This paper describes the unique challenges presented when using computer generated infrared imagery in a real-time hardware-in-the-loop simulation. IR target signature and background scene validation methods are discussed. The time delay introduced into the system when using a computer-generated image in a real-time closed loop simulation is discussed along with an analysis of the effect of the time delay on the hardware. Specific examples of the image and system analysis will be presented along with a visual comparison of 'live' and simulated IR sensor scenes.

In the past year, one measurement has been of great interest to those that we support and that is the measurement of optical turbulence or more specifically the refractive index structure coefficient. It is well understood that this parameters is affected by such factors as cloud cover, humidity, wind, solar loading and surface conditions. We present here field measurements from several locations where all of the environmental factors effecting turbulence have been measured along with the turbulence itself. This field data was compared with the US Army Atmospheric Sciences (ASL) Laboratory model called IMTURB. The results of this comparison are presented with comments on areas of improvement both in accuracy and in ease of use.

A critical review of rate constants for collisional relaxation processes and spontaneous emission processes in the non-LTE environment between 50 and 300 km is in progress. The infrared radiators considered are CO, H₂O, O₃, NO, CO₂, OH and CH₄. The last major compilation of vibrational relaxation rates relevant to the atmosphere was published by Taylor in 1974. The advent of laser-induced fluorescence techniques during the past two decades has greatly improved the state of knowledge regarding level specific collisional deactivation processes. Additionally, improved measurements of the rovibrational band strengths for H₂O, O₃ and CH₄ have resulted in improved Einstein A coefficients for these species. The rate constants under review are important parameters in models used to predict the dissipation of radiation entering the Earth's atmosphere as sunshine or energetic electrons. They are fundamental quantities in background radiance calculations. The effect of revising the rate coefficients has been illustrated using the Strategic High- Altitude Atmospheric Radiance Code (SHARC 2.0).

Measurements of the night sky spectral radiance were made with a Fourier Transform spectrometer at angles between 1 degree(s) and 90 degree(s) elevation over the wavelength region from 2.5 to 10.0 micrometers . The LOWTRAN 7 and MODTRAN computer models were used to predict the path radiance from ground to space for the same angles as the measurements. The comparison between the measured and predicted path radiance is shown for both models.

The emitted and reflected contributions to infrared sea radiance are affected by factors such as water temperature, prevailing weather conditions, wave structure, and viewing geometry. These factors are all addressed in the computer code SYNSEA which generates synthetic IR sea background imagery. In generating a radiance value for a scene pixel, SYNSEA calculates the radiance contributions for a distribution of surface orientations, weights these contributions according to the probabilities with which the orientations will occur, and sums the weighted contributions. SYNSEA predicts a unique surface orientation distribution for each pixel based upon the IFOV and the randomness of the surface, allowing the radiance calculations to make a smooth transition between the extremes of high and low resolution. SYNSEA has been coded in a modular fashion to facilitate modifications for integration with other codes. Although the statistics of the clutter predicted by SYNSEA have not yet been compared to any measurements, mean radiance values predicted with SYNSEA calculations have compared well with radiometric measurements.

This paper provides a new practical approximate solution to the radiative transfer equation under a narrow angle approximation with an anisotropic phase function. This assumption is useful for the case of optical and acoustic beam propagation in the ocean and the atmosphere, where particles sizes are much larger than wavelength. The two-scale expansion method that was found useful for statistical moments of wave propagating in continuum is extended to handle a general scattering kernel. Our procedure is applicable to other transport equations encountered in mathematical physics. Possible applications useful for remote sensing are briefly mentioned, and the complexity of computation is discussed.

The usual answer to estimating the atmospheric degradation of a forward looking infrared (FLIR) system's performance is to use an absorption, or better yet, an extinction coefficient derived from LOWTRAN calculations. The actual quantity that is appropriate, however, is target contrast transmission that includes system and source spectral characteristics, path radiance, and inherent contrast changes. Until now only transmissometer systems have ben used to measure and compare absorption and extinction values and predictions for imaging system performance. The Mobile Imaging Spectroscopy Laboratory of the U.S. Army Atmospheric Sciences Laboratory was used to measure directly the target contrast transmission by comparison of calibrated closeup and distant IR imagery of a uniform temperature large area target board. Only through proper choice of target and background area dimensions are the target contrast transmission measurement values consistent with LOWTRAN predictions. The important additional parameter that is too often overlooked is the modulation transfer function (MTF) of the atmosphere which can severely change the effective magnitude of the contrast transmission value that actually applies to a specific target feature or hot spot. Measurements are presented that illustrate how dramatic this interdependence of target contrast transmission and atmospheric MTF can be.

Mixing/shear layer turbulence is the major contributor to optical degradation effects experienced by a windowed hypersonic vehicle. A critical component in the prediction of these aero-optic effects, is the distribution, relative sizes, and velocities of the turbulent structures found within the mixing/shear layer. Previous attempts have had difficulty in measuring these high frequency, small scale turbulent properties. Therefore, a novel non-intrusive optical technique called the fiber optic flow monitor, was developed. This device was used in conjunction with a dual nozzle aero-optic simulator to experimentally determine turbulent flow properties and investigate their relationship to image distortion. The flow field studied was a dual species mixing/shear layer that had a mean flow velocity of approximately 430 m/s with a calculated mean turbule size of 0.7 mm. It was observed that the turbulent structures redistributed incident collimated energy into unique patterns of light. By monitoring these patterns, it was possible to measure several flow field properties. Data, gathered from this technique, was used to compute a statistical distribution of turbule velocities that was compared to theoretical predictions and image distortion parameters. Close correlation between experimental and theoretical values confirms that the technique provides a non- intrusive method of accurately characterizing small scale, high velocity turbulent structures.

A nonintrusive optical technique has been developed that measures the velocity components of high velocity mixing/shear layers. Within the high velocity turbulent media, flow structures exist that can randomly concentrate and/or redistribute incident light into unique intensity patterns. By observing the motion of these patterns over short time intervals, it is possible to deduce the velocity of the flow. A series of laboratory experiments was conducted to demonstrate the technique using the Teledyne Brown Engineering Dual Nozzle Aero-Optic Simulator (DNAOS). A binary gas, classical mixing/shear layer with a mean flow velocity of approximately 460 m/sec was generated for the tests. Two independent Q-switched Nd:YAG laser beams (1.064 micrometers ) were colinearly aligned, directed through the flow, and then recorded with a high speed CCD square array. Each laser was fired once during a camera frame, with a measured time delay of 1.36 microsecond(s) between the two laser pulses. The frames were taken at 92.5 Hz and stored for post-test analysis. By identifying the projected flow structure patterns and measuring the displacement of the patterns as recorded by the camera, two dimensional velocity components were calculated. These values were in fair agreement with mean flow velocities predicted with an empirical flow field prediction code.

This paper presents comparisons between LOWTRAN 7 and field measurements taken during a number of field tests conducted by the US ARMY Center for Night Vision and Electro- Optics at several different locations. The data was collected by the TECOM Ft Belvoir Meteorological Team. Data includes transmission at 0.4 - 0.7 microns, 1.54 microns, 3 - 5 microns, and 8 - 12 microns. A brief description of the equipment and meteorological parameters are presented along with the measured transmission and calculations using LOWTRAN 7 and in one example, MODTRAN.

A number of the experiments on the relationship between the angle amplification factor and the distortion of a night vision system and input illumination level is given in this paper. The causes for the changes of the amplification factor and,'or the distortion of a night vision system with the variation of input illumination level are discussed.The paper corrects a mistake in the related. technical documents of a night vision system and/or a night vision imaging device.

A loss of beam intensity occurs when a laser beam is propagated through shear layers with optical index fluctuations. A new model which uses Fraunhofer diffraction around a sinusoidal phase grating predicts the loss of intensity accurately. The size of coherent structures within the shear layers determines the spacing and the width of the grating. Numerical results agree well with previous experimental results.

Development of a computer-based system for generating simulated infrared imagery is described. The system provides realistic representations of infrared targets and backgrounds for training soldiers in combat vehicle identification (CVI). Using Army-supplied lists of desired items, Georgia Tech Research Institute (GTRI) constructed geometric and thermal models of NATO and Warsaw Pact combat vehicles, terrain backgrounds, countermeasures, distractors, and obscurants. Simulations of several fielded infrared sensors enable system users to generate training imagery sets, both snapshots and animated sequences, showing realistic sensor effects. The system is workstation-based and has a user interface that permits a non-expert to generate desired imagery sets from menus of available models and scenarios.

Camouflage design and evaluation requires that the developer assess the detectability of a camouflaged target in a natural, cluttered environment. Traditional techniques have involved panels of observers, but modern camoufleurs are seeking a satisfactory means of making reasonable estimates by the use of computer modeling. However, no adequate target detection model exists at this time. Camouflage experts at Belvoir Research, Development, and Engineering Center (BRDEC) are seeking means of improving the best of the current target detection models to make it capable of serving their needs. This is an interest which they share with many others within the Army who find that the best available model is unable to cope with low-observable targets. Unmodeled effects are responsible for the shortcomings. Current work at BRDEC and other organizations within the Army is examining such diverse methods as the use of fractals, edge evaluation, gray level cooccurrence matrices, power spectral analysis, and correlation lengths.

Recognition and identification ranges for targets in backgrounds are determined by the combined signatures of both target and background. In order to improve our understanding of the temporal behavior of cluttered backgrounds at infrared (IR) wavelengths in varying meteorological conditions, a series of experiments are described to model the observed IR imagery using the basic meteorological parameters recorded by a synoptic weather station. The acquired imagery covers a period of 44 hours. A simple multi-linear regression model using the basic meteorological data as input adequately describes the observations. Generally, only one to four weather parameters are required to model the observed apparent temperatures.

It is highly desirable for designers of automatic target recognizers (ATRs) to be able to test their algorithms on targets superimposed on a wide variety of background imagery. Background imagery in the infrared spectrum is expensive to gather from real sources, consequently, there is a need for accurate models for producing synthetic IR background imagery. We have developed a model for such imagery that will do the following: Given a real, infrared background image, generate another image, distinctly different from the one given, that has the same general visual characteristics as well as the first and second-order statistics of the original image. The proposed model consists of a finite impulse response (FIR) kernel convolved with an excitation function, and histogram modification applied to the final solution. A procedure for deriving the FIR kernel using a signal enhancement algorithm has been developed, and the histogram modification step is a simple memoryless nonlinear mapping that imposes the first order statistics of the original image onto the synthetic one, thus the overall model is a linear system cascaded with a memoryless nonlinearity. It has been found that the excitation function relates to the placement of features in the image, the FIR kernel controls the sharpness of the edges and the global spectrum of the image, and the histogram controls the basic coloration of the image. A drawback to this method of simulating IR backgrounds is that a database of actual background images must be collected in order to produce accurate FIR and histogram models. If this database must include images of all types of backgrounds obtained at all times of the day and all times of the year, the size of the database would be prohibitive. In this paper we propose improvements to the model described above that enable time-dependent modeling of the IR background. This approach can greatly reduce the number of actual IR backgrounds that are required to produce a sufficiently accurate mathematical model for synthesizing a similar IR background for different times of the day. Original and synthetic IR backgrounds will be presented. Previous research in simulating IR backgrounds was performed by Strenzwilk, et al., Botkin, et al., and Rapp. The most recent work of Strenzwilk, et al. was based on the use of one-dimensional ARMA models for synthesizing the images. Their results were able to retain the global statistical and spectral behavior of the original image, but the synthetic image was not visually very similar to the original. The research presented in this paper is the result of an attempt to improve upon their results, and represents a significant improvement in quality over previously obtained results.

In the development and analysis of sophisticated JR detection and recognition systems it is necessary to have a priori knowledge of the background clutter . In this paper, the spatial structure of ground-based infrared cloudy sky images is analyzed in terms of the spatial power spectrum and the spatial autocorrelation function . The experimental results we obtained for ground-based JR cloudy sky images do not fit the analytical model of the Wiener Spectrum 1,2 that is frequently used to describe natural clutter sources in the infrared .The spatial structure of ground-based cloudy sky IR images was found to be dependent on the percentage cloud cover in the image . Acorrected model was developed that relates the spatial structure of cloudy sky images to the percentage cloud cover in the image.

Detection of airborne point targets in infrared surveillance is hampered by the presence of cloud clutter. In order to develop tactical decision aids providing the performance of surveillance system in clutter conditions a scheme has been developed to obtain information on cloud clutter in real scenes. The amount of clutter depends on the amount of RMS variations of the intensity in the image within a certain cell size. An image consisting of RMS values in these cells give a first clutter map. To determine clutter for a large number of filter sizes it is necessary to apply the process for different cell sizes. These RMS values are weighed to obtain a single clutter value for a larger scene. In the process sensor noise needs to be separated from real background structure. The scheme has been applied to a number of 10 micrometers IR-18 image sequences for extensive analysis. After fine-tuning it is possible to compare the degree of clutter in different (types of) images. From the IR-18 image sequences several parameters have been determined. The temporal behavior of the images has been analyzed in two-dimensional scatter diagrams.

An analytic model of the clutter or variation of the IR surface radiance of ocean surfaces has been developed based on the Beckmann Spizzichino theory of scattering from rough surfaces. The model is cast in terms of an average bidirectional reflectance distribution function (BRDF) and contains the surface descriptions in terms of the average surface variance, sea surface correlation length, and Fresnel reflection coefficients. The average BRDF is then used to obtain an expression for the directional emissivity. Finally, the average sea surface directional radiance, as well as its variance, are obtained considering solar and sky reflections, as well as the surface thermal emission contribution.

The Army performance assessment methodology for visual and infrared countermeasures typically uses an area weighted average temperature/contrast (AWAT or AWAC) description of an equivalent uniform target in a uniform background scene. A simplistic target/background description is not sufficient to evaluate most signature countermeasure (CM) applications. This paper analyzes alternative strategies for using a realistic target/background model to achieve valid assessments of CM technology.

Camaeleon is a computer model developed by us for the assessment of camouflage using digital image processing techniques. The main topic of Camaeleon is a filter bank with bandpass-filters which are similar to the filters constituted by the receptive fields of the neurons in the early stages of the visual cortex. It is possible to obtain estimates of local energy (that is local contrast), local spatial frequency and local edge orientation from images processes by such filters. These features appear to be the most important concerning separation of figure and ground, that is object detection. The histograms of these features--for object and background separately--can be calculated and compared to obtain measures for similarity between object and background. At present, a functional description between calculated measures of similarity and detection probability derived from field studies is prepared. First results show sufficient correlation between similarity measures and detection probability for practical purposes of camouflage assessment and development.

A new technique is shown for refining and reducing incoming camouflage data based upon the Bayesian paradigm. Innovation is displayed in use of a statistical conditioning sequence that avoids the need to form target features from the data. The result is a simplified and more accurate probabilistic indication of actual target presence. This probabilistic indication can then be incorporated into a variety of target detection scenarios or, alternately, to form the basis of a theoretically optimal Bayesian target detector. Numeric simulation is presented to show the effectiveness of the technique against simulated camouflage.

An application of image analysis techniques for construction of a field-portable, computer- based system for evaluation of personal camouflage is presented. The scope of evaluation is limited to target acquisition reduction performance of camouflaged uniforms against the unaided human visual system. Evaluation of camouflage with this system is based on a set of features for each component in the CIELAB color space that includes statistical texture measures, and first-order descriptive statistics over specified regions in a scene as functions of orientation and scale. The approaches taken for data acquisition, reduction, processing and evaluation are presented. An analysis of camouflage data relative to a human panel test is also presented.

It is well known that CO₂ laser reflectances of natural surfaces containing certain minerals, notably quartz, feldspar and kaolinite, exhibit differential reflectance features in the 9 - 11 micrometers mid-infrared spectral range. The use of off-normal reflectance ratios using measurements at four CO₂ laser wavelengths to differentiate between various types of soil surfaces has been established. Off-normal reflectance ratios are observed to be relatively independent of incidence angle compared to ratios computed at normal incidence, which makes them suitable for field remote sensing applications. Road surface materials, such as concrete and asphalt, contain large quantities of quartz, and as such exhibit reflectance characteristics similar to soils. Our measurements indicate that it is, nevertheless, possible to discriminate between road and soil surfaces using off-normal reflectance ratios. Discrimination from soil surfaces is better for concrete compared to asphalt.

This paper presents a sea radiance polarization model and experimental measurements of near- horizon sea glint polarization in the 3 - 5 micrometers and 8 - 12 micrometers spectral bands. The experimental measurements include the effects of polarization on the glint statistics, the degree of linear polarization and the polarization signal-to-noise ratio (SNR) improvement factor for both spectral bands in the presence of sea glint. The results indicate that the polarization in the 3 - 5 micrometers spectral band is dominated by the reflected solar and sky radiance and is polarized in the s plane. The polarization of intense sea glint in the 8 - 12 micrometers region is low and s polarized due to the weak solar spectrum in this band. In little or no glint, the radiation is weakly p polarized. Experimental data indicate that a polarizing filter can produce a significantly larger SNR improvement for the 3 - 5 micrometers spectral band than for the 8 - 12 micrometers band. Theoretical calculations using the polarization model show good agreement with the experimental data.

The authors report the statistical analysis of a digitized cluttered background scene containing a military ground vehicle. This is the first phase of a study to evaluate several sensors and scenes to generate a statistical measure of IR sensor performance based on the pixel by pixel correlation of the output imagery.

Longwave infrared imagery of cloud fields are examined in terms of their power spectral density (PSD). In order to systematically investigate the dependence of the PSD on viewing conditions, a cloud scene simulator is employed to generate images of a simulated cloud field. The cloud field is fully three dimensional and is describes by its fluctuating temperature and liquid/ice water content fields. The imaging process accurately calculates the spatially varying attenuation of blackbody emission. Several views of a single cloud field are examined to study the effect of viewing angle on the image PSD. Zenith views produce isotropic PSDs, while nearly horizontal views contain a large amount of foreshortening and a correspondingly anisotropic PSD. One possible component of the foreshortening is simply geometric and can be estimated and compared to the simulation output. We find that geometrically induced foreshortening does not describe the PSD effects observed in the simulation for the relatively thin cirrus-like cloud simulated here. Possibly this indicates that the three-dimensional cloud structure is more important in some views that in others when there are large fluctuations in the cloud optical properties. We are pursuing a more quantitative description of this behavior.

CI Systems has made progress in the quest for ever more realistic simulation of infrared scenes for testing of advanced missiles. This has been achieved by building a completely automatic PC controlled electro-optical system which reproduces a textured object, moving with respect to a background, as it is seen by an approaching missile with a 3 to 5 (mu) FLIR. An independently moving and approaching countermeasure is also present in the scenario. The missile approach is simulated by a 10:1 infrared zoom; the radiance texture of both the object and the background are achieved by an infrared transparency, whose pixel radiance can be chosen from among 256 gray levels.

Until recently, the source-based photometric scale of NIST had been derived from the gold- point black-body using a long derivation chain resulting in a scale uncertainty of about 1%. In order to improve the accuracy of the photometric scale, a new detector-based illuminance scale has been realized with an uncertainty about twice as small as that of the previous source- based scale. The realization of the new illuminance scale was made by photometers calibrated for absolute spectral response, traceable to NIST primary standard detectors using a short derivation chain. The luminous intensity scale is derived from the illuminance scale. With the introduction of the new photometric scale, the photometric calibration services of NIST will include luminous intensity and luminance calibration of different kinds of light sources, measured with standard photometers. Luminous intensity measurements of LEDs are discussed as an example to illustrate the new calibration services. In addition, the new photometric scale will allow calibration of illuminance and luminance meters with standard photometers, using different kinds of light sources.