A resonant cavity electro-optic frequency modulator at 2.0 GHz was developed for high frequency heterodyne characterization of HgCdTe de-tectors. The standard technique for heterodyne evaluation of HgCdTe detectors beats a CO₂ laser with the filtered incoherent output of a black-body radiator. For detectors which will be incorporated into laser radar systems, it is desirable to perform characterization tests with a coherent, narrow band signal. The electro-optic modulation technique represents a quick, precise measurement that uses a signal which closely simulates the laser radar Doppler shifted return signal. In the electro-optic modulator, a 2.0 GHz microwave signal is coupled into a CdTe crystal to fre-quency modulate a CO₂ laser beam. A dielectrically-loaded cavity insures the phase velocity match of the microwave and the IR signal. The design also allows for modulation at higher order resonances, thus the HgCdTe detectors can also be characterized at 4.0 GHz. No such electro-optic modulator, operating in the GHz frequency range for the evaluation of high speed photodetector performance. is known to have been reported to date.

This paper presents the design and test performance of a conical cavity type blackbody radiance source that will meet the requirements of the Halogen Occultation Experiment (HALOE) on the NASA Upper Atmospheric Research Satellite program (UARS). Since a radiance source meeting the requirements of this experiment was unavailable in the commercial market, a development effort was undertaken by the HALOE Project. The blackbody radiance source operates in vacuum at 1300 K + 0.5 K over any 15-minute interval, uses less than 7.5 watts of power, maintains a 49°C outer case temperature, and fits within the 2.5 x 2.5 x 3.0 inch envelope allocated inside the HALOE instrument. Also, the unit operates in air, during ground testing of the HALOE instrument, where it uses 17 watts of power with an outer case temperature of 66°C. The thrust of this design effort was to minimize the heat losses, in order to keep the power usage under 7.5 watts, and to minimize the amount of silica in the materials. Silica in the presence of the platinum heater winding used in this design would cause the platinum to erode, changing the operating temperature set-point. The design required the development of fabrication techniques which would provide very small, close tolerance parts from extremely difficult-to-machine materials. Also, a space rated ceramic core and unique, low thermal conductance, ceramic-to-metal joint was developed, tested and incorporated in this design. The completed flight qualification hardware has undergone performance, environmental and life testing. The design configuration and test results are discussed in detail in this paper.

High efficiency glass polarizers have been developed that offer superior design characteristics over plastic, wire-grid and some prism type polarizers. Glass polarizers have transmittance greater than 90%, contrast greater than 500:1 at acceptance angles up to 60 degrees, and excellent resistance to humidity and temperature. High polarizing efficiency is obtained over a bandwidth of about 300nm. The location of the polarized band can be specified within the near-infrared range (800-1800nm). The polarizing effect is due to the resonant absorption of elongated sub-microscopic particles of silver metal aligned along a common axis in the glass. The polarizing mechanism is discussed briefly followed by a review of typical glass polarizer performance, design characteristics and suggested applications.

A monolithic 128 x 128 InSb array is described for staring infrared imaging systems operating in the 3-5μm spectral region. The array is fabricated with only 4 mask levels and has almost 5 times higher responsivity and nearly 14 times greater wafer yield as compared to a previous design. The higher responsivity has resulted in demonstration of significantly improved thermal imagery.

Commercial avalanche photodiodes (APDs) have been operated as single-photon detectors at an optimum operating temperature and bias voltage. These detectors were found to be 1.5 to 3 times more sensitive than presently-available photomultiplier tubes (PMTs). Both single-photon detection probability and detector noise increase with bias voltage; detection probabilities greater than 25% were obtained with detector noise levels comparable to the noise of a PMT; higher probabilities were measured at higher noise levels. The sources of noise and their dependence on temperature and bias voltage are discussed.

A review is made of infrared (IR) sensor technologies for spaceborne surveillance applications. Particular emphasis will be made on the status of short wavelength IR (SWIR), medium wavelength IR (MWIR) and long wavelength IR (LWIR) materials, detector arrays and readout electronics, specially developed for acquisition and tracking. Included in the review are extrinsic silicon detectors for space surveillance, Pt/Si, InSb, PbS, HgCdTe and PbSe for missile surveillance, planar hybrid and Z-geometry hybrid focal plane architectures for configuring large IR sensors, and scene generator technology for sensor verification and validation.

Optical systems have become a paramount ingredient in the Strategic Defense Initiative (SDI) Architectures. Their functions and usefulness is illustrated in a hypothetical element of a far-term space based system.

This paper looks at the processing of multidimensional bandlimited signals which have been sampled on hexagonal lattices. The hexagonal sampling theorem and its attendant aliasing are reviewed and algorithms for digital filtering and interpolation between hexagonal and rectangular lattices are presented. The efficiency of these procedures is discussed. The hexagonal lattice is shown to possess some definite advantages compared to the rectangular lattice with respect to computational complexity, particularly for isotropic systems. These advantages, however, probably do not outweigh the need for interpolation on data which has already been sampled on a rectangular lattice.

Analytical results are presented showing improved detection and location performance in a staring sensor using a hexagonal detector array or having the capability to adjust the size of its blur spot (variable focus). Square and hexagonal detector arrays are compared for detection probability versus focus, for location accuracy versus focus and signal-to-noise ratio, and for location accuracy versus blur size and detector array fill factor.

Innovative detector sampling and sensor scanning concepts are presented for reducing the signal bandwidths in non-imaging infrared sensors. The use of these concepts can significantly reduce the data rates in the interconnect cables between the detector focal plane and its associated signal conditioning circuits. The data rate reduction is achieved with only a minimal loss of performance.

The image of a circular source aperture is convolved with a Gaussian representing the random errors in the optics. The convolution is Fourier transformed. The transform of the aperture image is approximated as a Gaussian. The approximation introduces little error into the inverse Fourier transform which returns the convolution to real space. Thus, the spot diameter is given by: (Total Random and b = Image Decoll 4a )1a² + b² where a = Decoll 4σ Diam ) ' where a is the characteristic radius of the Gaussian and 4o is the diameter which encloses 86.5% of the incident light flux. The approximation introduces less than 2% error in the (4σ) total decollimation. The approximation is generalized to include diffraction effects due to the diameter of the optics.

The most critical element in the 20-year evolution of infrared search and track (IRST) systems has been the development of acceptable signal processing techniques. Acceptable, in this case, refers to both the ability to operate in a wide variety of changing background and target/engagement conditions and the ability to operate reliably with minimum insertion losses under benign conditions. Tactical and strategic military applications for endoatmospheric operations have tended to emphasize scanning IRST designs to meet the wide field of view and rapid update requirements with available detector technology. The need to automatically detect targets at long ranges usually leads to the adaptation of spatial processing techniques to extract point source targets from extended background clutter. Usually a combination of processing techniques are required to provide the performance and reliability necessary to achieve long range detections and low false alarm rates (typically one per hour or less). This paper identifies the generic classes of signal processing concepts that have been applied to endoatmospheric military applications. It reviews the evolution of these concepts and provides examples of currently popular techniques and their relative performance with emphasis on initial target declarations.

Conventional algorithms for predicting the position of a maneuvering target are based upon the point-mass properties of the target; e.g., the position and velocity of the "center-of-gravity". In addition to providing estimates of these tradi-tional target states, an IR imaging sensor can be used to make inferences regarding the "fine structure" of the target. For example, the orientation of the target with respect to the image plane has a major impact on the perceived tracking dynamics. It follows, therefore, that an algorithm which utilizes orientation information will tend to make smaller mean-square prediction errors than does one which is based upon point-mass states alone. This paper presents a image-based algorithm for predicting the position of an evasive target moving in space or the upper atmosphere where the aerodynamic drag is small. This paper employs a simple stochastic model of acceleration and orientation dynamics to derive a useful prediction algorithm. This prediction has an easily computed form even in the presence of interfeature dependence. The paper concludes with a simple example illustrating the procedures proposed in the paper. The relationship of this approach with a "constant-velocity' extrapolation is made explicit in the example.

This paper presents a mathematical error analysis for jitter compensation of an imaging sensor. First, a model is developed for the traditional approach to jitter estimation. Next, a general mathematical result is derived to evaluate the efficacy of the polynomial model for gradient estimation. Finally, an approximate error analysis is presented for the accuracy of individual terms in the polynomial model.

The pseudoregistration approach to moving target Identification (MTI) is discussed. Even though techniques based on linear interpolation have proven to be adequate for present generation sensors, higher order interpolation will be needed for future sensors. A higher order polynomial interpolation approach is discussed and shown to be robust in the face of random noise as long as data windows of sufficient length are used. A problem of data "extrapolation" which has been encountered in its implementation is illustrated along with the current solution of "frame exclusion." This problem is caused by the displacement of the current image frame being different than the displacements of image frames used for the interpolation.

Improvements in infra-red (IR) sensor sensitivity have resulted in background clutter becoming a limiting factor for target detection. The extreme IR clutter intensities which occur following detonation of nuclear weapons is a major limitation to target detection capabilities. This paper investigates improvements in the signal-to-clutter ratio provided by simple two-dimensional spatial filtering of IR detector array data. The difference in the spectra for natural and nuclear background clutter require different filter parameters to provide optimal performance.

Schottky-barrier focal plane arrays are the only infrared imagers that are fabricated by the well established silicon VLSI process. Therefore, at the present time they repre-sent the most mature technology for large-area high-density focal plane arrays for many SWIR (1 to 3 μm) and MWIR (3 to 5 μm) applications.^1-27 - PtSi Schottkybarrier detectors (SBDs) were developed for operation in the MWIR band at a temperature of 77 to 80K. These SBDs can be designed for operation at 77K with a dark current density in the range of 1.0 to 20 nA/cm². Pd₂Si SBDs were developed for operation with passive cooling at 120K in the SWIR band. Although the PtSi SBDs have rather low quantum efficiency (0.5 to 1.0% at 4.0 μm), however, because of very low readout noise, the IR-CCD imagers with PtSi SBDs are capable of 300K thermal imaging with a noise equivalent (NEAT) of less than 0.1K. The noise floor of the SBD arrays is limited by the SBD dark current shot noise.

We have produced highly efficient planar and binary optical elements such as lenses, beam profile shapers, beam multiplexers, and coherent laser beam adders. A binary relief pattern on these devices controls the phase and amplitude of transmitted or reflected wavefronts. To fabricate these devices, we bring together developments in EM binary grating theory, in high resolution E-beam lithography, and in reactive ion-beam etching techniques. We demonstrated this technology by producing diffraction-limited lenses on thin planar substrates with >96% diffraction efficiency. These lenses can be used in monochromatic and wideband FLIR telescopic applications. These devices are also useful as high-speed scanner elements, and play important roles in coherently adding laser beams from modular or monolithic laser arrays.

Since the first SPIE conference session on imaging spectroscopy in February 1981, the development of several significant instruments and the initiation of yet larger programs has taken place. The Near Infrared Mapping Spectrometer for the Galileo mission to Jupiter has been completed and awaits a launch opportunity. An advanced version of this instrument has been selected for the Mars Observer mission and is a candidate for several additional planetary missions. An Airborne Imaging Spectrometer has completed several seasons of data collecting in the United States, Australia, and Europe. An advanced instrument for NASA's U-2/ER-2 high-altitude research aircraft is nearing completion. Significant design work has been accomplished on the Shuttle Imaging Spectrometer Experiment and design studies for NASA's Earth Observation System include a major imaging spectrometer instrument. Researchers in a variety of disciplines have reported significant results from aircraft instrument data and the development of sophisticated data analysis software for manipulating and interpreting the high dimensionality data sets is progressing. Progress in this growing technique of passive remote sensing is contributing to the planning and design of future instrumentation.

Infrared imaging sensors are required to provide visual target recognition at night. The potential exists for converting present day-only imaging sensors to day/night by replacing the vidicons or silicon charge-coupled devices with an infrared staring focal plane array. The analyses show that twice the recognition range can be obtained with a 3 to 5 micron sensor as compared to an 8 to 12 micron sensor for an aperture-limited system. This paper describes the performance of antiaircraft fire-control imaging sensors based on a staring 128 x 128-element indium antimonide detector array operating in the 3 to 5 micron Medium Wavelength IR band. With a 4.5 inch diameter clear aperture telescope, head-on recognition of fighter-size aircraft (for example, Foxbat) is predicted to be as far as 16 nautical miles, with detection at over 100 nautical miles. A laboratory sensor has been used to accumulate nighttime imagery of commercial aircraft at 8 to 15 mile ranges.* The results are presented in a short video tape in which the aircraft can be easily recognized.

The millimeter-wave dielectric properties of a series of IR window materials were determined over the temperature range 23-1000°C. Materials studied included Al₂0₃, ZnS, ZnSe, aluminum oxynitride (ALON), and magnesium-spinel (MgAl₂0 ₄). These materials all exhibited fairly high millimeter-wave dielectric constants (e' = 8-10), but with essen-tially negligible room-temperature losses for most applications. However, both the dielectric constant and loss tangent increase significantly with increasing temperatures. The increases in dielectric constant with temperature can be analyzed in terms of a macroscopic dielectric virial expansion model, and are primarily due to the effective increase in volume for each polarizable unit of the material. Consequently, a strategy to overcome this degradation would be to search for new materials or composite structures with low thermal expansion coefficients. The observed millimeter-wave loss properties are characteristic of contributions from intergranular impurities and show an onset of increased absorption at ca. 500°C. However, even at 1000°C, typical loss tangents are still below 0.05, and should be acceptable in most millimeter-wave window applications for reasonable thicknesses (/ < X).

The contract requirement for the LRIRSS program was to design, fabricate, test, and deliver infrared (IR) surveillance systems capable of target detection and recognition at extended ranges. Several significant technical advancements were made during the course of the program. The twenty three inch primary reflecting surface had to be of aluminum due to thermal and weight considerations. This was accomplished through the use of diamond point turned aluminum. The IR and laser receivers are active simultaneously; this has been achieved through the incorporation of a "chopping" mirror that time shares the main aperture between the FLIR and CO₂ laser receivers. In addition, the laser had to be eye-safe at the transmitter exit aperture. The use of a single pulse Carbon Dioxide (CO₂) laser has met the requirement, and is certified by the US Army Environmental Hygiene Agency in Study No. 25-42-0335-85. The major technical challenges have been met, as well as several minor difficulties that arose during the execution of the effort. All hardware has been delivered to the Government in accord with the contractual schedule. NOMENCLATURE CO2 Carbon Dioxide CPU Central Processing Unit FLIR Forward Looking Infrared FOV Field of View IFOV Instantaneous Field of View IR Infrared LRF Laser Rangefinder LRIRSS Long Range Infrared Surveillance System PCB Printed Circuit Board

This paper describes the Automatic Target Recognizer Working Group. Included are a brief history of the organization, its goals, and progress to date. Information on the organization and meetings is included. Five committees are described--data base, evaluation, applications, artificial intelligence, and security.

Data representative of imaging sensors and scenarios which form the inputs for automatic target recognizers (ATRs) is critical to their development, testing and performance evaluation. The Data Base Committee of the Automatic Target Recognizer Working Group provides a forum and produces products to assist collection, distribution and use of data for development of military ATR systems. Examples discussed in the paper include digital image data exchange format specifications. Requirements for ground and image truth data have been the subject of surveys. Such inputs are intended as recommendations for consideration by imagery data collection activities whose products are potentially useful for ATR development. Other topics concerning collection, reduction, use and exchange of imaging sensor data are outlined but not discussed in detail.

The evolution of automatic target recognizers (ATRs) and their application to military problems has necessitated the development of new techniques for the evaluation of these devices. The evaluation of today's ATR must provide some measure of the expected perfor-mance in a real scenario. It is essential that the evaluation techniques evolve at the same rate as the ATR technology and possess the same relative maturity. The Automatic Target Recognizer Working Group (ATRWG) Evaluation Committee was formed as a forum for the exchange of ideas and to lead in the development of ATR performance measures. This paper presents the standards and procedures developed by the DOD and accepted by ATRWG's Evaluation Committee. Also included is a brief history of the evaluation of modern ATRs. Definitions of performance as well as testing procedures are outlined that are uniform, accurate, extendable, and usable for evaluation of ATRs. A scientific methodology is being developed for evaluating improvements in targeting effectiveness for various DOD missions and scenarios.

The objective of the ATRWG Applications committee is to bridge the gap between users and developers of ATR technology. This is being done by developing ATR specifications for relevant DoD missions. This paper will describe the work being done by the Applications committee. Specifically, this paper will contain a description of a generic ATR specification format, a brief discussion of strawman specifications for diverse mission areas and a bogus example of a specific mission area. These draft strawman specifications will be presented to DoD user commands and project managers offices to assure that the ATR developer has formulated ATR performance objectives and requirements that are realistic in light of today's technology and are properly addressing user needs.

The ATRWG Security Committee participates in the development of Automatic Target Recognizers (ATRs) and image processing guidelines as related to TEMPEST requirements and policies set forth by the Department of Defense (DoD).

"First-generation" ATRs represented the first, concerted attempts by Government and industry to automatically exploit digital FLIR (forward looking infrared) imagery. However, these ATRs suffered fundamental limitations which rendered them unacceptable for operational use. The incorporation of AI (artificial intelligence) in ATR designs has the potential to overcome these limitations. The AI Committee of the ATRWG (Automatic Target Recognizer Working Group) is educating the ATR community on AI technology and evaluating the utility of various AI techniques in ATRs.

An infrared target recognizer system usually consists of the infrared sensor, the target recognizes: processor and the target recognizer algorithms. Initially the problem was detined that sensors existed to support automatic targeting and the risk was in the algorithms and processor development. The current perspective is that the fielded generation of infrared sensors do not support fully-functional target recognizer requirements. It is also believed that the current algorithm approaches may be successful given "correct" sensor inputs. This paper discusses the second generation of FLIR sensors and the system impacts on automatic target recognizers.

The United States Navy's worldwide operations on land, at sea, and in the air require it to be capable of performing a wide variety of missions. This paper will attempt to define how Automatic Target Recognition (ATR) technology, as it relates to infrared (IR) systems, is currently being used, where it will be used in the future, and what the perception of ATR technology in the Navy is. Rather than try to cover all possible Navy missions which are relevant to ATR technology, this paper addresses three specific missions in the area of airborne IR weapon systems. These areas which are currently receiving research and development emphasis in the Navy include IR Ship Classification, Long Range Detection of Non-Resolvable IR Aircraft Signatures, and Ground Target Cueing.

A unified program is proposed which blends modification of Army tactics with research and development to obtain short and long term goals for technological application of ATRs for improving weapon systems. A pragmatic approach of understanding present technology while carrying out research on analytic methods is required. The keystone of both these programs becomes development and testing of real time processors both in the field and laboratory.

Automatic target recognizers (ATRs) have been under development for the last 10 to 15 years and have apparently failed to-date to live up to the great expectations of the Department of the Defense (DoD) community. In this paper, I will try to show that the poor press being generated about ATR performance results in recent government field tests should not be misinterpreted as a blanket indictment of the entire ATR development community.

Automatic target recognition using state-of-the-art imaging sensors requires a combination of several related research areas in information processing: image feature extraction, pattern recognition, and artificial intelligence. The state of the art in these areas has advanced significantly in the last decade. Advances have also been made in the development of active and passive imaging sensors. These developments, combined with the advances in computers and microelectronics, enable the development of an Automatic Target Recognizer (ATR). However, there are no fully operational ATR's deployed today. The emphasis in ATR system development and testing has been on hardware rather than algorithms for information processing; also, more funds have been expended on sensors than on proces-sors. Overall sensor/processor system considerations, cost/performance trade-offs to match the mission requirements, and susceptibility to countermeasures have frequently been neglected. This paper points out critical technical issues for ATR's and explains Northrop's approach and activities in ATR development.

This paper discusses past, present and future strategic aircraft requirements for ingress and egress, then focuses on the tech-nologies of the CO₂ Laser Radar and the Automatic Target Recognizer. Present systems currently consist of a mix of various sensors which are not correlated until each is presented to the operator. Additionally, active sensors are highly detectable by threat warning systems, while passive sensors do not provide critical range information. CO₂ Laser and AIR tech-nologies will significantly contribute to the resolution of these issues.

Texas Instruments feels that the development of Automatic Target Recognition (ATR) technology offers the potential for substantial performance improvements in future weapons systems. The ability to automatically find and launch weapons against targets in military environments will relieve attack aircraft or helicopter pilots of a demanding assignment in a high-stress environment, thereby improving their effectiveness and survivability. Future guided missiles can have ATRs aboard and thereby have the ability to autonomously find their own targets. In that case, launch platforms will be required only to transport a missile to a suitable launch position, and a pilot need not locate the target before launching the missiles. ATR implementation requires three areas of development: sensors, signal processing algorithms, and processing architectures. The number and type of sensors required to fulfill an ATR requirement are dictated by the mission to be performed. The processing architectures which will implement the signal processing algorithms will consist of numeric, parallel, and symbolic processing elements.

The purpose of this paper is to explore fundamental considerations surrounding an ATR's algorithm design process. Emphasis is placed on a breakdown of the problem into the constituent elements that are deemed most important in affecting ultimate operational success. They involve both data base concerns and key algorithmic subset interactions.