The use of diffractive optical elements ( DOE ) in most refractive reimaging infrared optical systems significantly simplifies the optical design form and improves the image quality. The basic theory of chromatic aberration correction using DOE is analyzed. A 3 to 5 um design is shown comparing the optical design form and image quality of the conventional design and the DOE improved design. The DOE improves performance while lowering cost and weight by reducing the number of lens elements and desensitizing the design to misalignments. MTF, MRT, and NEDT tests of the hardware empirically demonstrate the superior performance of the DOE design.
The D.C. level of the detector data for the U.S. Common Module FLIR is lost due to A.C. coupling of the individual detector elements in the Common Module electronics. In this paper, a method is presented for recovering the relative D.C. level between detectors. This method has been implemented and demonstrated using a GFE Thermal Imaging Set (TIS) from the Gunner's Primary Sight of the M1A1 ABRAMS Tank.
A technique is presented for compensating detector gain and offset variations that is applicable to both scanning and staring sensors. The method requires that the sensor line of sight be moved in a regular fashion over the scene to provide data for the compensation. The method does not depend on the scene characteristics. Equations describing the method and establishing its performance are presented. An implementation approach is described. The effects of scene dynamics are discussed.
This paper describes a procedure useful for registering images generated by second generation FLIR sensors with images produced by existing low resolution sensors. The method is particularly applicable to the problem of aligning the seeker of a precision guided munition, such as a missile, to a narrow field of view targeting FUR. A robust image processing technique for measuring the residual registration error from sensors with dissimilar resolutions was developed and tested. The specific development presented in this paper concerns the alignment of an image generated by a second generation targeting FLIR with a field of view of 2 x 2.75 degrees sampled by a pixel matrix of 480 x 640 elements to another sensor's image with a field of view of 1.5 x 1.5 degrees sampled by a pixel matrix of 128 x 128 elements. Although the method addresses sensor offsets, it can be extended to determine scale as well as rotational differences. The algorithms were developed using video imagery collected with a second generation FUR prototype built by Westinghouse. Algorithms were developed and tested using state of the art image processing software running in a graphics workstation environment.
This paper is the third and last in a series dealing with spatial and directional noise and its effect on FLIR performance. First a simplified version of the FLIR90 MRTD model is explained. Then it is shown that the characteristics of various 3-D noise elements can be used to develop the correction functions kh(f) and k,v(f) which adjust the horizontal and vertical MRTD upward to account for additional system noise not accounted for by the conventional NETD. The current version of FLIR90 uses kh(f) and kv(f) evaluated at f0 and used as constant adjustments kx and ky. FLIR92, to be released early in 1992, will use the full frequency dependent functions kh and k, This method of MRTD prediction using a correction process to account for 3D noise is validated using several examples covering second generation FLIRs and staring systems.
Acquiring a human target through an electro-optical sensor is of great interest to a variety of governmental agencies. However, relatively little field data collection had been focused in this area, so that there remained some uncertainty in the correct choice of modeling parameters. The Night Vision range performance modeling methodology can accommodate this unique target in addition to the more conventional tactical vehicle. Recent experiments at the NVEOD in human target detection have provided insight into the task involved and the system modeling requirements. This paper recounts the results of two experiments involving personnel target detection with FLIRs. The first experiment involved first generation TOW night sights in a ground-to-ground scenario and the second experiment involved two different airborne FLIRs, one parallel and one serial scan, in an air-to-ground scenario. In both experiments the targets included walking and standing men. The ACQUIRE model was exercised against the test results to determine the appropriate values to be used for the characteristic dimension and task cycle criteria for moving and stationary personnel targets in a non-search scenario. Additional theoretical discussion in the paper explores static performance and search concepts in relation to motion and detection.
Two-point calibration is often used to correct for nonuniformities across focal plane arrays (FPAs), as well as for calibration. Because the input-output curves of FPA channels are nonlinear, two-point calibration produces a systematic calibration error as a function of flux, and the channel-to-channel variations of this calibration error leave a significant post-correction nonuniformity. A physical model of detector nonlinearity is used to illustrate these points. A simple formula is proposed, which fits the input-output curves much better than the straight line used by two-point calibration, and is almost as easy to use. When the new formula is used, the system's performance is no longer sensitive to the choice of calibration temperatures, and no longer degrades rapidly outside the calibration interval.
In the last two years, Hughes Aircraft Company and the Santa Barbara Research Center have demonstrated that LWIR PV HgCdTe staring focal plane arrays can be fabricated reproducibly with high performance and high yield. Record yields for detectors and readouts, in excess of 50%, have been achieved on 128 x 128 arrays with 40 x 40 gm pixels on 40 jam centers. Key to the successes are the development of p-on-n PV HgCdTe detector technology and high density CMOS readout circuitry. In 1988, Hughes Missile Systems Group established an inventory account whose goal was to fabricate 128 x 128 LWIR FPAs for a broad range of missile requirements. They range from high background tactical applications to low background SDI scenarios. The staring FPA chosen couples a high impedance p-on-n LWIR PV HgCdTe detector array to a CMOS high capacity direct injection readout array and is designated the DI-128. The detector arrays were fabricated with cutoff wavelengths ranging from 9.2 to 9.9 gm and with RoAo products ranging from 100 to 1000 S2-cm2. The excellent RoAo products allow a simple direct injection input circuit to be used while maintaining injection efficiencies in excess of 80% allowing near BLIP performance. The DI input circuit has a charge handling capacity in excess of 20 million carriers. A variable integration time capability is provided for dynamic range management and performance optimization. 39 LWIR HgCdTe FPAs have been fabricated by the inventory account to date with a yield of nearly 50 %. The arrays were tested at 77 Kelvin with an f/2.0 aperture at 295 Kelvin in the 8.0-9.0 gm spectral band resulting in a background flux of 6.0 x 1015 photons/cm2-sec. The array average NEAT achieved was typically in the range 0.025 to 0.020 Kelvin. Excellent dc and ac uniformities of 4-6% were universally observed. The yielded FPAs all had greater than 98.5% operability with many parts achieving greater than 99.5%. Additional tests were performed to determine if the LWIR detectors could be used in the MWIR. The NEAT achieved in these tests was approximately 0.025 Kelvin. the outstanding detector RoAo products achieved at SF RC reduce the leakage current sufficiently to achieve this performance level in the MWIR. The excellent MWIR performance opens the possibility of simultaneous MWIR and LWIR imagery using a single LWIR staring FPA.
Hybrid HgCdTe 256 x 256 MWIR focal plane arrays have been developed to meet the sensitivity, resolution, and field-of-view requirements of advanced infrared imaging systems. The best devices have RMS response nonuniformities less than 3% of the mean. They have been characterized over the temperature range 80-160K. For tactical infrared backgrounds, the FPA is photon noise limited at operating temperatures up to 130K With standard two-point correction, spatial fixed pattern noise is reduced below the background photon noise. Detector arrays are fabricated on sapphire substrates using the PACE-1 (Producible Alternative to CdTe for Epitaxy) process. PACE material growth techniques have been improved to optimize detector yield, uniformity, and performance in the MWIR band. The readouts in the hybrid FPAs are CMOS switched-FET silicon integrated circuits processed with 1.5 arm design rules. The charge capacity of each 40 pm x 40 pm unit cell is 4 x 107 electrons. The video output is capable of data rates exceeding 10 MHz.
A hybrid 640 x 480 PACE HgCdTe FPA is being developed to meet the needs of many applications including missile seekers and FLIR's. The device will offer full TV resolution with sensitivity much superior to PtSi, having over one-quarter million pixels. The hybrid is comprised of a PACE HgCdTe detector array having nominal 5 pm cut- off wavelength, mated to a high speed CMOS readout having high charge-handling capacity. The HgCdTe detectors are being fabricated on alternative sub- strates using Rockwell's mature PACE-1 and advanced PACE-3 detector growth processes. The PACE-3 process, which is currently being developed, involves the MOCVD of HgCdTe on buffered silicon sub- strates. Since a PACE-3 hybrid uses silicon substrates for both the readout and detector, excellent hybrid reliability is expected even after thousands of thermal cycles. Based on 2562 results to date, the PACE-3 detectors will have mean quantum efficiency < 30% with rms respons- ivity nonuniformity < 8% across a typical array. The mean detector RoA product measured on parallel test samples is typically greater than 104 sl-cm2 at 80K. PACE-1 detectors in this cell size have RoA < 105 o-cm2 at 80K and quantum efficiency < 60%. The 640 x 480 readout uses a CMOS switched-FET architecture with direct injection input. The device is structured in four quadrants and has four high speed outputs. The unit cell size is 27 x 27 um 2. Included in the cell is a built-in test feature allowing full readout characterization prior to hybridization. Maximum data rate of each output is < 10 MHz. The 640 x 480 thus generates over 109 bits/sec of video information. The chip was fabricated using 1.25 um lithography at a silicon foundry. Very good functional yield of the < 3 cm2 dice was achieved.
An overview is presented of linear drive cryo- coolers under development at Hughes Aircraft. A family of linear coolers is discussed ranging from miniature linear coolers (designed for man-portable and vehicle mounted systems) to ultra-long life linear coolers designed for spaceborne applications. Performance char- acteristics of these linear coolers are discussed with an emphasis on demonstrated reliability. Results of exten- sive reliability testing on coolers and key components are detailed, demonstrating the life characteristics of these coolers. A discussion is presented on the successful integration of a linear cooler into a system incorporating focal plane array (FPA) detectors. Key performance characteristics of this cooler that contributed to the successful demonstration of this system are delineated. The status of linear cooler technology is sum- marized, indicating that reliability in excess of 4,000 hours MTTF has been demonstrated.
Two-point nonuniformity correction has been realized at Amber Engineering using an analog technique for the correction of pixel offsets and gains in real time. This system has been designed for use with Amber's 128 x 128 focal plane arrays, and a new system based upon similar concepts is being built for Amber's 256 x 256 arrays. The analog circuitry is much smaller and consumes less power than digital circuitry performing the same operations. Better than an order of magnitude reduction in electronics volume as well as power dissipation has been realized using this analog technique as opposed to the digital method while maintaining the same level of performance. Extrapolation of these techniques leads to incorporation of nonuniformity correction on FPAs in a generation of readout devices that will be available in 1992.
Low contrast details in wide-dynamic-range IR imagery may well be lost in global, monotonic map- pings from raw signal levels to 8-bit displays. Local contrast enhancement techniques can bring out such detail through mappings which are no longer global or monotonic. Three groups of algorithms for this purpose are described and compared with regard to degree of enhancement, artifacts, and computa- tional costs. The groups are: local implementations of histogram projection; "modulo" processing, which is a triangular-shaped many-to-one mapping; and high frequency enhancement as implemented in the spatial domain but designed in the spatial frequency domain. Results are illustrated by a representa- tive subset of the numerous IR images surveyed.
This paper presents a description of a modular simulation architecture which has been implemented on a 386 personal computer ruining UNIX and X-Windows to model virtually any staring or scanning array system and its associated processing electronics. The modularity of the simulation system permits rapid software-prototyping of a proposed sensor system and provides the designer with a valuable "what if' tool for evaluating the impact of numerous design alternatives on the performance of the overall system. Such front-end simulations performed during the early stages of a system-level design are effective in reducing costly design iterations and they help to ensure first-run system success.
This analytical study focused on the range perfor- mance impact of FPA charge handling capacity (CHC) and residual nonuniformity after compensation (NUC). It was found that the CHC and (especially) NUC characteristics of an IR sensor are at least as important as--and often more important than--traditional parameters such as quantum efficiency, spectral band, F/number and/or fill factor. These results have significant implications for IR sensor system design tradeoffs incuding those allocating FPA "real estate" between active area, charge storage wells and readout circuits--a trade made more complicated because readout circuit linearity is, a factor in estab- lishing FPA uniformity. The merits of hybridization to achieve high fill factor must be questioned unless non- uniformity is adequately addressed as well. It is clear that the entire sensor system--including CHC, NUC and the Nonuniformity Compensation subsystem--must be considered in selecting the "best" sensor configuration.
Modern imaging sensors incorporate complex focal plane architectures and sophisticated post-detector processing. These advanced technical characteristics create the potential for multi-component noise generation which can exhibit effects temporally as well as along the vertical and horizontal image directions. Such complex three-dimensional (time, vertical, horizontal) noise cannot be adequately treated by previous mathematical analyses developed for simpler system designs where detector noise was predominant. In a parallel sense, earlier methods for noise measurement are no longer satisfactory. A new methodology has been developed at C2NVE0 to characterize the noise patterns exhibited by advanced thermal imaging systems. This effort has the ultimate objective of improved system noise modeling in the Center's FLIR90 thermal performance model. In the new methodology the total noise present in the output of the sensor is divided into an set of eight independent components. Each of these components is characterized by the presence or absence of specific characteristics in each direction in a three-dimensional coordinate system. The method represents a significant expansion of the standard techniques to characterize thermal system noise. This paper, the first in a series of three, explains the principles behind the 3-D noise methodology and the methods used. A second paper published simultaneously will describe how this methodology is implemented in a laboratory measurement environment. A third publication planned for a future date will discuss and apply the modeling techniques being developed to incorporate this noise methodology in FLIR90.
A two color 64 x 64 staring HgCdTe infrared sensor was used to characterize the effects of background flux and focal plane temperature on system uniformity. The sensor consists of an infrared focal plane, Cassegrain telescope, and a 14 bit processor to provide offset and gain calibration and a 12 bit digital output. The data were taken on consecutive frames of video frame rates of 100 Hz. The twelve bit sensor output was stored on a high data rate recorder/reproducer allowing a slower data rate for the lower speed test station, thus allowing analysis of consecutive complete frames of data. This paper merely identifies the effects of these changing parameters - the analysis of the physical mechanisms which result in these characteristics and the modeling to be able to predict these effects will be conducted over the next year.
Historically, Forward Looking InfraRed (FLIR) imaging systems have undersampled because of the small number of scanned detectors in the focal plane and because the human visualizing process can compensate for aliasing in the image. The advent of larger, higher density arrays of detectors, as well as the sensitivity of computerized Automatic Target Recognizer (ATR) computers to aliasing effects has made the study of spatial and temporal sampling effects a high priority. This paper describes a test bed designed to explore the effects of varying the optical and electrical parameters of line-scanning and staring focal plane arrays (FPAs) on viewed images and on ATR algorithm effectiveness. Microscanning staring and scanning telescope optical systems are employed. Electrical support electronics to operate various types of LWIR and MWIR focal plane arrays and test software to acquire and reduce data about them are described. A real-time imaging system which can reorganize up to 16 FPA outputs into an image up to 1024 by 1024 pixels allows the effect of varying parameters to be observed directly. The system also includes a massively parallel processor which will execute various ATR algorithms on the FPA images.
Measurement of infrared radiation for the detection of air turbulence such as Low Altitude Wind Shear (LAWS) and high altitude Clear Air Turbulence (CAT) is pursued. CO2 infrared absorption and re-emission in the 13p to 16p region is exploited. A laboratory sensor instrument is being developed with cryogenically cooled mercury cadmium telluride linear array in the scanning mode with multiple spectral filters to cover the 8p to 12p transmission band as well as the 13p to 16p CO2 absorption band. A future option of utilizing an uncooled planar pyroelectric array in the staring mode is contemplated. Current effort is concentrated on developing the instrument as well as analysis and simulation. It is anticipated that a field demonstration unit will be available in the summer of 1991.
This paper had its origin during an informal discussion between the authors and Ken Ando at the National IRIS last summer. The discussion was on the importance of displays to Thermal Imaging System Performance. We concluded that it would be worthwhile to hold a workshop on the subject, bringing together the various disciplines involved, to share information and exchange views on the subject.
Emphasis is often placed on the sample rate or spacing of sampled imagery, with less attention paid to how the output is constructed using the samples. The fidelity of a sampling process, however, depends on the reconstruction method as well as on the sample rate. This paper will discuss the benefits of display processing to improve image quality in sampled sensors. Topics will include: basic sampling theory, the application of Fourier Transforms to sampled imaging systems, the origin of phase artifacts in sampled imagery, and a discussion of the benefits of display processing when using more display pixels than sensor samples. Examples will be shown of images generated using different reconstruction techniques.
In the modeling of first generation (Common Module) FLIR performance the effects of obvious focal plane sampling in the vertical (cross-scan) direction were neglected. That methodology used the analog (unsampled) horizontal system quality exclusively as the measure of overall performance. In spite of this, the models were successful because the FLIR focal plane standardization in effect at that time resulted in a relatively fixed relationship between horizontal system quality and overall system performance. Therefore, there was little incentive to investigate the effects of sampling. With the emergence of advanced thermal imagers, the situation in regard to sampling has changed completely. These systems are inherently digital in design and as a result introduce image sampling in both the horizontal and vertical directions. In particular, the imagery from staring focal plane thermal imagers is thoroughly embedded with sampling characteristics since, in the basic system designs, they are limited to one sample per detector spacing in each direction. The experimental study described in this paper was motivated by the need to determine the specific effects of sampling on FLIR performance for these advanced systems, and the extent to which other system characteristics such as pre-filtering and post-filtering MTF's modify the direct effects of sampling. A number of perception experiments have been performed at C2NVE0 wherein sampling, pre-filtering, and post-filtering parameters were varied systematically and the effects on military observer identification and recognition performance were measured. The results of these experiments are being analyzed for incorporation in the Night Vision Advanced FLIR Performance Model.
Laboratory assessment of Forward Looking Infrared (FLIR) devices has long been the standard method for characterizing the performance of first generation scanning thermal imagers. The correlation between test and field performance has been widely documented. With the continued insurgence of discretely sampled staring array systems it is necessary to evaluate the need and validity of these tests as they pertain to staring array systems. This paper will discuss the relevance of the standard tests with respect to staring array measurements and show the results of these measurements. The paper will also look into possible new ways of evaluating staring arrays that will be of more relevance to those people who are attempting to model the field performance of the staring array. It is hoped that these results will lead to updated test methodology that will more closely relate to field performance.
Proc. SPIE 2075, Phenomenology/sensitivity analysis of platinum silicide, indium antimonide, and mercury-cadmium-telluride detector performance in the 3 to 5 micrometer infrared band for air-to-ground applications, 20750P (29 January 1992); doi: 10.1117/12.2300253
Platinum silicide, indium antinomide, and mercury cadmium telluride are compared as detector materials for an air-to-ground reconnaissance/targeting sensor using the 3 to 5 micrometer infrared band. The comparison is based on a scaled count rate involving only the detector material quantum efficiency, the reflected and thermal emission from the target and background, and the atmospheric trans- mission--all of which are highly wavelength dependent. Last year, the results of analyzing baseline scenario and seven sensitivity scenarios were presented. In these analyses, only the target was considered in the independent variation of each of the following parameters: target temperature, target reflectivity (emissivity), atmospheric visibility, atmospheric aerosol type, and humidity. This year, in a follow-on effort, several baseline scenarios were constructed to determine relative contrast between the standard target and four backgrounds--vegetation, soil, sand, and snow. Six sensitivity analysis scenarios were constructed in which target and background temperatures are varied from the baseline. In three additional scenarios, the atmospheric humidity and target reflectiv- ity are independently varied from the baseline. Initial studies to explore the benefit of dual band operation within 3 to 5 micrometers have been done.
The Army requirement to fly helicopters at low level at night led to the development and fielding of night vision pilotage sensors. These sensors have included image intensifiers (I2) operating in the near infrared as well as 8 to 12 micron thermal imagers. The design of current pilotage sensors was driven by available technology. There were no clear data for optimum pilotage sensor design, to enable the designer to trade off sensor field of view (FOV) and resolution or to predict the performance increase which could be obtained by increasing sensitivity. The current Center for Night Vision and Electro-Optics (CCNVEO) effort is an attempt to establish design criteria for night pilotage sensors. It includes flight experiments to find sensor characteristics which optimize flight tasks as well as assessments of the performance of fielded systems. We conclude that terrain flight can be accomplished with reasonable pilot workload using a sensor with 40 degree FOV and 0.6 cycles per milli- radian (cy/mrad) resolution. Larger FOV or better resolution will lesson workload and improve confidence; however, the ability to resolve scene detail of 0.6 cy/mrad is essential and should not be traded for increased FOV. Further, a pilotage system which provides both thermal and I imagery will significantly enhance system capability to support a variety of flight tasks under a wide range of environments. We also conclude that solid state cameras with detector dwell time equal to the standard video field rate are not suitable for use in pilotage systems. The long dwell time leads to image blur due to the head and scene motion associated with many pilotage tasks.
We are investigating a temperature evaluated mine position survey (TEMPS) for remote detection of buried land mines. The TEMPS methodology uses two passive IR channels peaked near 5 and 10 microns to decouple temperature frbm emissivity related effects. The true (corrected) temperature maps show surface temperature variations of 0.2Ã‚Â°C. Corrections are made for air-path interference and reflected sky radiation. We exploit a property of Planck's radiation law which applies for small temperature excursions from 288 K. The radiant emittance is proportional to emissivity times absolute temperature to the power of (50/wavelength in microns). Our corrected temperature maps show patterns of conducted heat generated by buried objects which heat and cool at different rates than the surrounding materials. These patterns are distinguished from the patterns produced by surface objects. Their respective spatial, spectral, thermal, emissivity and temporal signatures differ. EG&G flew a dual-band IR scanner at 60 m for our demonstration of the TEMPS methodology at Nellis AFB. We detected simulated mine targets covered by 10 cm of dry sand. Optimization of this technology is expected to enhance the capabilities of the military community for standoff mine detection and other applications.