When modeling thermal imaging systems to predict their performance, there are three functional areas that must be specifically characterized: the object being imaged; the intervening atmosphere; and the sensor itself. This paper will discuss two computer simulations presently being used by the U.S. Army Missile Command (MICOM) which integrate models of these three functional areas to calculate the performance of a variety of thermal imaging systems for a range of atmospheres, scenarios, targets, and sensors. Common to both approaches is: use of the Johnson bar criteria for detection, recognition and identification; the use (direct or indirect) of the AFGL LOWTRAN code to model the atmosphere; and, finally, some methodologies for modeling staring focal plane arrays (FPAs) as the sensor detection element. The three functional areas will be discussed first, followed by discussions of the two approaches; the Fire Control Sensor Simulator (FCSS), an engagement model; and, the MICOM Infrared Imaging System Performance Model (MIISPM), a more detailed model emphasizing the sensor itself. The MIISPM is based upon the Night Vision Laboratory Static Performance Model (NVLSPM), but has been enhanced in several important ways, discussed in the paper. While both approaches provide as output the same performance criteria (probability of detection, recognition and identification, as a function of range), each model has its own strengths for different aspects of the sensor performance problem: for the FCSS, targets and scenarios; for the MIISPM, details of the sensor. Both approaches will be presented and compared in terms of the results, run time efficiency, and required hardware. Both models presently run on IBM or IBM compatible personal computers.
An evaluation of various infrared imaging systems was performed to determine their abilities to identify thermal anomalies in buildings. The systems were tested under environmental temperatures from -20°C to 25°C for their minimum resolvable temperature differences (MRTD) at spatial frequencies between 0.03 to 0.25 cy/mrad. The temperature dependence of MRTD was analyzed and compared with the predicted values in ASHRAE standard 101-83 for thermal imaging systems. The temperature dependence of infrared systems' object temperature calibrations was investigated. The signal transfer function (SiTF) of infrared sensors were generated to verify and calibrate the dynamic range of each sensor. Also discussed are the results of measurements of modulation transfer function (MTF) of infrared imaging systems, which are based on Fourier Transforms of the line spread function (LSF). It is shown that the results of the MTF calculations can be correlated with their MRTD measurements.
Minimum Resolvable Temperature Difference (MRTD) is an invaluable tool in designing and testing infrared sensors and is to a lesser extent valuable in predicting field performance. The history of MRTD use is briefly reviewed and some shortcomings of MRTD are noted especially in the discrepancies between predicted and tested performance in the laboratory. Newer developments in infrared technology have altered the form of sensors from the FLIR concepts originally treated by the existing MRTD models. Infrared Linescanners, Focal Plane Arrays and other second generation sensors must be included. Some suggestions are given for modifying existing MRTD models to include these other sensors. A concept of "objective MRTD" is presented to allow use of MRTD in systems such as Automatic Target Screeners (ATS) and Automated Test Equipment (ATE) where the human operator is temporarily bypassed.
Calibrated thermal imagers are being used more frequently than non-imaging radiometers to gather target and background infrared signature data. Current radiometric calibration guidelines should be expanded to include tests specifically designed for imagers. This paper recommends several tests for this purpose and presents data collected on an Inframetrics 210 as an illustration of the value of the tests.
For nearly five years, the Air Force Avionics Laboratory has operated the Targeting Systems Characterization Facility (TSCF) at Wright-Patterson Air Force Base, Ohio. This has resulted in an extremely large amount of archived data describing the performance of an infrared sensor, the target and backgrounds observed by the sensor, and the environmental conditions occurring during the measurement periods. Since these data are stored in a digital computer in a rigid format, they can be efficiently and conveniently accessed. The TSCF data are available to the sensor community and have proven to be a valuable asset in the development and validation of the Research Grade Infrared Tactical Decision Aid, and in the investigation of environmental influences on infrared sensor performance. This paper describes the Targeting Systems Characterization Facility itself and demonstrates the usefulness of the data. Emphasis is placed on the measurement procedures and instrumentation used for the data collection. The preparation of this paper is sponsored by the Air Force Avionics Laboratory.
This paper presents characterization data for a platinum silicide, staring mode, infrared camera. The camera focal plane is an interline transfer CCD with a 244 X 160 array of PtSi photodiodes. We also present samples of the imagery obtained with this IR camera. Our results demonstrate that the performance of silicide based infrared sensors is now superior to industrial thermal imaging systems that are based upon scanning.
Characterization of fixed-pattern noise is an important aspect in evaluating the performance of a staring array. The usual method of quantifying the effect of this artifact on the image is to state the variance of the pixel levels. However, this implicitly assumes that the fixed-pattern noise is spatially "white", that is, it has an equal effect at all spatial frequencies of interest. Usually, fixed-pattern noise has a nonrandom spatial distribution, which violates the assumption of white noise. A more complete characterization is provided by the spatial power spectrum of the fixed-pattern noise. This descriptor quantifies the effect of fixed-pattern noise on image data, both in terms of its frequency content, as well as its magnitude. Consideration of the noise spectrum is seen to yield additional insight into the nature of the fixed-pattern noise present on the array.
Modeling of Schottky barrier focal plane arrays (FPAs) is relatively new as is the tactical use of these arrays. Only limited data sets are available for model validation. Available data have been used to develop and validate an MRT model including noise calculations. The primary area of concern is the MRT equation, which is also true for scanning sensors. However, it is believed that the causes for discrepancies between the model and the data are different from those for scanning sensors. This paper provides a brief description of Schottky barrier FPAs and their potential tactical applications, and then discusses some of the available performance data. The model description shows calculated performance compared to measured performance including a discussion of questionable areas.
The NPL hemispherical reflectometer is used between 2.5 and 55 micrometres to measure spectral and total emissivities for the calculation of radiative energy transfer for materials with a uniform surface. However, materials used in energy saving and military applications are often non-uniform with random patchiness or with a repeating pattern or intrinsic structure. A new facility has been devised at NPL for the measurement of the spatial distribution of emissivity of non-uniform materials. A Rank Taylor Hobson TICM-II infrared thermal imager is used to study the variation of radiance across the surface of a sample maintained at a uniform near-ambient temperature. Reflected radiation from the sample is eliminated by covering most of the hemisphere of sample irradiation with a cold non-reflecting screen. A 2.7x telescope supplied by Rank Pullin Controls together with a new deep-focus special attachment, developed by NPL, allows samples to be viewed at distances from infinity down to 0.27 m. The resulting images are enhanced by digital filtering in a GEMS customised image processing system with a framestore holding 4 images of 512 x 512 pixels at 16-bit depth. This provides a highly flexible system for the measurement of the emissivity of any element of a sample from 100 mm down to a fraction of a millimetre across. A range of statistical, analytical and graphics functions is available from the GEMS software. Corrected emissivity maps are derived from comparison with uniform standard samples, measured on the NPL hemispherical reflectometer and residual non-uniformities are corrected using large NPL reference black body cavities.
A parallel scan forward looking infrared (FLIR) with a digital scan converter (DSC) electronically reformats parallel input analog data into serial digital output data. This digital data can be supplied directly to a subsystem for further processing. Prior to returning to the analog domain to create the video image, the digital data passes through a line-to-line interpolation operator, gamma correction look up table and a histogram processor. Because of the video processing, the characteristics of the analog signal different than the digital signal. Line-to-line interpolation can provide √2 signal-to-noise (SNR) improvement. Frame integration, depending upon the scheme used provides additional improvement to the SNR. Clearly, the video image may not be representative of what actual signals are being supplied to the subsystem. Since the characteristics are different, it is essential to measure the system intensity transfer function (SITF), and noise equivalent temperature difference (NEDT) in both the analog and digital domains. Appropriate digital data display (histogram of values) permits easy assessment of data quality. Examples of missing data bits (dead lines) are shown. Lower order missing bits can affect a subsystem but the effect is minimized on the display due to the line-to-line interpolation scheme. The NEDT, SITF and noise spectral density in the digital and analog domains are compared. The implications of the difference in both domains are discussed.
Inter-frame operations are a very powerful, yet neglected, class of image processing operations in which the operands are entire frames, not individual pixels as in intra-frame processing. For inter-frame processing the processed frames must be registered. A procedure for the automatic registration of images of both two and three dimensional object scenes is described. Hardware implementation of this generic process, exploiting recent developments in processing hardware technology, would make available this class of powerful image processing operations for a wide range of military and industrial applications.
Software has been developed for the simulation of laser radar range imagery. Two versions have been developed: the first is an idealized model which is noise-free and with zero dropout rate; the second includes both pointing and range noise effects and provides for calculation of probability of detection for each pixel, with dropout occurring for probabilities below threshold, and also allows for user control over a number of other parameters such as scanning convention (unidirectional vs. bidirectional), scan efficiency, and trajectory update rates. Each version allows for motion of a LADAR sensor across a terrain database on which faceted objects (targets and clutter) have been placed. For each pixel the program calculates the laser exit beam direction, based upon the combined effects of the sensor sweep pattern and the motion and attitude of the sensor platform. The exit beam is traced for intersection with the terrain or an object. Program output consists of the x,y,z-coordinates of the intersection point and the (real-number) range to that point for each pixel. This output can then be converted to a displayable range image. The software is currently implemented on a VAX 11/750 computer operating under VMS.
Infrared signature data on a number of unique military targets have recently been acquired under a limited variety of environmental conditions. A method has been developed for extrapolating these signature data to other environmental conditions without resorting to heat transfer modeling. The only environmental factors considered are solar loading, sky radiation, wind speed and air temperature, and vehicle operational state. Coarse faceted geometric models of the targets are constructed from a very limited number of material types whose responses to environmental factors are determined from the data. Projections for an armoured target are compared with the predictions of two signature models which utilize more detailed and sophisticated approaches.
A sophisticated and flexible digital infrared data collection system has been assembled at the Georgia Tech Research Institute. The system includes such components as two radiometers responding in the 3-5 and 8-12 micron spectral bands, a near infrared television, and a master computer which performs the data acquisition and permanent storage function. In addition, a second computer system is included allowing for off-line image analysis. The application software development and hardware interfacing comprised the majority of the effort described herein. The purpose of the system is to allow for the collection of infrared target signature data which hopefully will enhance development of phenomenological and semi-empirical thermal models. This paper presents a description of the instrumentation hardware and software as well as details on the interface between sensors, computers, and additional peripherals. The details of a recent system characterization activity are also included.
This paper describes parallel-scan thermal imagers based on cooled MCT detectors operating in the 8- to 12-μm band. A line of unclassified commercially available imagers are shown. Equations for predicting the detection and recognition ranges are given, based on the U.S. Army Night Vision Laborator (NVL) performance model.
Thermal Imaging Systems have found innumerable applications in industry, research, medicine and the military. One of the most valuable applications is in non-contact temperature and radiation measurements. This paper explores the unique problems that are encountered when making thermal image measurements in the field. An approach to solving these problems by the separation of the field data acquisition process from the lab data analysis is presented. Desired features for each of the three instruments involved in this process; the radiometer, data recorder, and data analyzer, are discussed and examples of typical applications are shown.
Equipment description and preminary results of the field evaluation of an Automatic Infrared Test and Inspection System (AITIS) developed under a U.S. Air Force Manufacturing Technology contract sponsored by the Air Force Wright Aeronautical Laboratories are presented. The AITIS equipment can be used in repair depots and in circuit board manufacturing production facilities to test printed circuit board assemblies to detect and isolate circuit faults to the componment level through a noncontact infrared screening process.
Pyroelectric Vidicons have been commercially available for approximately fifteen years. Approximately 10 years ago I.S.I. Group, Inc. began manufacturing a television camera system incorporating the Pyroelectric Vidicon tube. Until this time most systems had been custom fabricated, for a specific application. The early camera systems were large, bulky, and difficult to use. These initial camera designs were updated to make them more user oriented and to generally simplify their operation. The original Thermal Imaging systems were comprised of a camera head, (which incorporates the tube), a camera control unit, (which incorporates all the necessary control electronics), separated by a camera control cable. These systems found many different applications, but had been restricted to areas where 110 Volt, 60Hz AC power was available. Within the past few years a new camera system has been developed. This camera system is completely portable, battery powered, and light weight. This camera systems now permits thermal analysis in areas that are remotely located. These camera systems produce images which are displayed on standard television equipment. No coolant is required to cool the detector. These camera systems respond only to thermal energy radiated by an object and no signal is produced by visible light.
A black-and-white infrared (0.9 to 2.2 μm) video camera, filtered to record radiation within the 1.45 to 2.0 μm mid-infrared water absorption region, was evaluated with ground and aerial studies. Imagery of single leaves of seven plant species (four succulent; three nonsucculent) showed that succulent leaves were easily distinguishable from nonsucculent leaves. Spectrophotometric leaf reflectance measurements made over the 1.45 to 2.0 μm confirmed the imagery results. Ground-based video recordings also showed that severely drought-stressed buffelgrass (Cenchrus ciliaris L.) plants were distinguishable from the nonstressed and moderately stressed plants. Moreover, the camera provided airborne imagery that clearly differentiated between irrigated and nonirrigated grass plots. Due to the lower radiation intensity in the mid-infrared spectral region and the low sensitivity response of the camera's tube, these video images were not as sharp as those obtained by visible or visible/near-infrared sensitive video cameras. Nevertheless, our results showed that a video camera with mid-infrared sensitivity has potential for use in remote sensing research and applications.