Since the 1950s, numerous studies have been performed within the surveillance and reconnaissance (S&R) and target acquisition (TA) communities in an attempt to predict information extraction performance as a function of image collection and quality parameters. In general, the work followed two separate paths. The TA community developed models to predict probabilities of detection, recognition, and identification as a function of target size, range, and collection system design/performance parameters (e.g., MRT, FLIR92,NVTHERM,MRC). The S&R community developed models to predict National Imagery Interpretability Ratings (NIIRS) as a function of system design and collection parameters (e.g. IR GIQE). More recently, efforts have linked the two approaches such that NIIRS can be predicted from TA models and probabilities of identification can be predicted from NIIRS. With both approaches, resolution is a dominant term. A considerable amount of variability and uncertainty results from target differences. The criteria used to define the NIIRS generalize target type, size, and level of identification specificity. The TA predictions use the Johnson recognition criteria to relate lines on the target to recognition performance. A recent paper found that TA predictions differed substantially between the visible and IR. Further, the paper reported substantial differences among vehicles in terms of a confusion matrix. This finding was not surprising in light of other research, but suggested the need for a more detailed examination and explanation of results. Accordingly, the current effort was undertaken. Data from a variety of past studies dealing with target recognition were examined relative to the Johnson criteria, along with a more detailed analysis of data from two recent TA studies. A hypothesis of target recognition performance was generated and partially validated using available data.

Experiments and analysis were used to determine the number of resolvable cycles across an alphanumeric character required for readability. This has serious implications for the resolution needed for a surveillance camera to present a “readable” image to a human. Fourier analysis was used to predict the number of cycles required for readability. Using two-dimensional Fourier transforms, the set of 26 English letters and 10 Arabic numerals was analyzed and classified. This theory is supported by empirical data based on user identification of random English letters and Arabic numerals. The results strongly indicate that accurate readability (defined as 90% correctness or better) can be accomplished with approximately 2.8 cycles across a block letter. This appears to suggest a lower resolution requirement than that generally accepted for unknown target identification. The reason is the limited data set of only 36 alphanumeric characters, of which the observer possesses a priori knowledge. Moreover, the ability to read an alphanumeric is a steep function of the resolution between 2 and 3 cycles per character height. The probability of correct “Reading” can be expressed similarly to that of Detection, Recognition and Identification by using a postscript such as “Read₉₀”.

A new target acquisition metric has been developed which provides better accuracy than the Johnson criteria. Further, unlike the Johnson criteria, the new metric can be applied directly to sampled imagers and to imagers which exhibit colored (spectrally weighted) noise or frequency boost. This paper describes the new metric. Experimental data are provided showing the problems with the Johnson criteria and illustrating the robust performance of the new metric.

In preceding work, it was shown that the relative error in the predicted intensity of an individual pixel in any broadband MWIR image simulation that employs some form of band-average emissivity and/or average detector responsivity approximation in its models will be about equal to the fractional standard deviation of the MWIR emissivity of the corresponding material. This relationship between the error in simulated integrated image intensity and the variation in spectral emissivity over the MWIR band for a set of 27 commonly encountered scenery materials was shown to behave like a simple power curve, with the power inversely proportional to scene temperature. However, what is more important than the error in a single simulated image pixel is how a multi-pixel simulated image is effected. In this follow-on paper, the errors associated with band-average emissivity approximations are quantified with respect to errors in synthetic images. Comparisons of image contrast and image correlation using band-average emissivity and spectral emissivity are performed. The impact of band-average emissivity in a simple synthetic image on a perception experiment is presented as an example of an application-dependent effect.

A summary of the development and impact of the Night Vision and Electronic Sensors Directorate (NVESD) Time-Limited Search (TLS) Model for target detection is presented. This model was developed to better represent the search behavior when an observer is placed under time-constrained conditions. The three primary components of the search process methodology are (1) the average detection time (based on characteristics of the image), (2) occurrence and time delay associated with false alarms, and (3) the time spent searching a Field-of-View (FOV) before moving on to another FOV. The results of four independent perception experiments served as the basis for this methodology. The experiments, which were conducted by NVESD, portrayed time limited search conditions for different sensor resolution and background clutter levels. The results of the experiments showed that these factors influence the search process and their impacts are represented within the components of the TLS methodology. The discussion presents the problems with the current model and details the constraints that must be understood to correctly apply the new model.

This paper describes a spectral night illumination model. The model provides spectral irradiance on a horizontal surface for wavelengths between 0.4 and 2.0 microns. These wavelengths encompass the visible, near infrared, and short wave infrared spectral bands. The primary significance of this model is that consistent estimates of spectral irradiance are now provided for the visible through SWIR spectral bands. The primary sources of night illumination are described. The paper also describes how the new model was derived from spectroscopic data gathered by astronomers. Model predictions are compared to standard references commonly used to predict night illuminations.

For an IR (infrared) sensor, the raw digital images coming out from the FPA (focal plane array) A/D converter contain strong non-uniformity/fixed pattern noise (FPN) as well as permanent and blinking dead pixels. Before performing the target detection and tracking functions, these raw images are processed by a CWF (chopper-wheel-free) MBPF NUC (Measurement-Based-Parametric-Fitting Non-Uniformity Correction) system to replace the dead pixels and to remove or reduce the FPN, as shown in Figure 1. The input to MBPF NUC is RIM_i,j(the raw image), where 1⩽i, j⩽256, and the output is CIM_i,j, the corrected image. It is important to note that as shown in Figure 1 the IT (integration time) for the FPA input capacitors is a critical parameter to control the sensor's sensitivity and temperature DR (dynamic range). From the results of our FPN measurement, the STD (standard deviation) of FPN from a raw uncorrected image can be as high as 300-400 counts. This high count FPN will severely reduce the sensor's sensitivity (we would like to detect a weak target as low as a couple of counts) and hamper the target tracking and/or ATR functions because of the high counts FPN artifacts. Therefore, the major purpose of the NUC system is to reduce FPN for early target detection, and the secondary purpose is to reduce FPN artifacts for reliable target tracking and ATR.

This paper describes a major generalization of a recently reported radiometrically-accurate algebraic nonuniformity correction (NUC) algorithm. The original technique was capable of accurately estimating the bias nonuniformity from a sequence of pairs of images exhibiting strictly one-dimensional (1D) subpixel shifts. The new technique relaxes the subpixel 1D shift constraint to arbitrary two-dimensional (2D) motion, which can be either sub-pixel or super-pixel. The 2D technique relies on calibrating only rows and columns on the perimeter of the array, which in turn, provides the algorithm with the necessary initial conditions to recursively estimate the bias values in the entire array. In this way, radiometric NUC can be achieved non-disruptively, as needed, without disturbing the functionality of the interior array elements. The 2D algorithm is highly localized in time and space lending itself to near real-time implementation. Radiometric NUC can be achieved with a relatively low number of frames (typically about 10 frame pairs). Moreover, as in its earlier 1D version, the performance of the 2D algorithm is shown to be insensitive to spatial diversity in the scene. This paper will address the performance of the 2D technique using real infrared data.

In tactical sensor imagery there always is a need for less noise, higher dynamic range and more resolution. Although recent developments lead to better and better Focal Plane Array (FPA) camera systems, modern infrared FPA camera system are still hindered by non-uniformities, a limited signal-to-noise ratio and a limited spatial resolution. The current availability of fast and inexpensive digital electronics allows the use of advanced real-time signal processing to address the need for better image quality. We will present results of signal-conditioning algorithms, which achieve significant better performance with regard to the FPA problems given above. Scene-Based Non-Uniformity Correction (SBNUC) can provide an on-line correction of existing and evolving fixed-pattern noise. Dynamic Super Resolution (DSR) improves the signal-to-noise ratio, while simultaneously improving spatial resolution. The signal-conditioning algorithms can handle camera movements, high temporal noise levels, high fixed-pattern noise levels and large moving objects. The Local Adaptive Contrast Enhancement (LACE) algorithm does effectively compress the 10, 12 or 14 bits dynamic range of the corrected imagery towards a 6 to 8 bits dynamic range for the display system, without the loss of image details. In this process, it aims at keeping all information in the original image visible. We will show that the SBNUC, DSR, mosaic generation, and LACE can be integrated in a very natural way resulting in excellent all-round performance of the signal-conditioning suite. We will demonstrate the application of SBNUC, DSR, Mosaicking and LACE for various imaging systems, showing significant improvement of the image quality for several imaging conditions.

Recent technology advances have made low cost, eye safe, high performance laser-range-gated (LRG) imagers a reality. These advances include the Electron Bombarded CCD sensor and the 1.5 micron, monoblock laser. LRG imagers use a laser beam to illuminate targets at extended ranges; the targets are then identified with the EBCCD sensor. Several features of LRG imagers make predicting range performance different than for passive imagers. LRG imagers are described. The features that make active imager performance different from passive imager performance are discussed. Features unique to active imagers include laser speckle in the image, the narrow illumination beam and its interaction with the atmosphere, the highly directional “spot light” illumination of the target, and the range gating of the receiver. This paper discusses the unique modeling requirements for LRG imagers.

We are developing a technique for accurately measuring spectral responsivity functions of infrared cameras using tunable lasers. We present preliminary results for uniform scenes where tunable infrared lasers illuminate an integrating sphere, diffusing the light to fill the imaging system optics. A commercial camera based on a liquid nitrogen-cooled InSb focal plane array was tested in the 1.4 micrometer to 4.7 micrometer spectral range using a continuously-tunable periodically-poled lithium niobate (PPLN) optical parametric oscillator. Another commercial camera based on an uncooled microbolometer array was tested using a discrete-tunable CO₂ laser in the 9 micrometer to 11 micrometer spectral range. Results from these tests show that signal-to-noise ratio, uniformity, stability, and other characteristics are favorable for use of this technique in the characterization of infrared imaging systems. We also propose a generalization of this technique, to include scenes with arbitrary, controlled spatial content such as bar patterns or even pictures, by illuminating a commercially-available digital micromirror device (DMD). Dependence on irradiance level, exposure time, and polarization can also be measured. This technique has an inherent advantage over thermal-emitter based methods in that it measures absolute spectral responsivity directly without requiring knowledge of the spectral emissivity or temperature of the source.

In designing a high performance IR re-imaging optical system for thermal imaging handheld applications, one of the most changeling tasks is to achieve very compact and light weight optical configuration, and at the same, be able to incorporate a number of associated optical devices, such as micro-scanner mirrors, non uniformity correction device, dual FOV optical elements, focussing lens etc.., within the optical path. This paper describes the First Order Optics calculation for such optical configuration, which allows optics engineer to define a suitable optical layout before performing optical optimisation and analysis. A design example of a 3-5um FPA re-imaging optics system is presented.

The non-uniform response in infrared focal plane array (IRFPA) detectors produces corrupted images with a fixed-pattern noise. In this paper we present an enhanced adaptive scene-based non-uniformity correction (NUC) technique. The method simultaneously estimates detector's parameters and performs the non-uniformity compensation using a neural network approach. In addition, the proposed method doesn't make any assumption on the kind or amount of non-uniformity presented on the raw data. The strength and robustness of the proposed method relies in avoiding the presence of ghosting artifacts through the use of optimization techniques in the parameter estimation learning process, such as: momentum, regularization, and adaptive learning rate. The proposed method has been tested with video sequences of simulated and real infrared data taken with an InSb IRFPA, reaching high correction levels, reducing the fixed pattern noise, decreasing the ghosting, and obtaining an effective frame by frame adaptive estimation of each detector's gain and offset.

FLIR technology continues to expand at a near exponential rate. Even with the rapidly changing FLIR technology the ways in which we evaluate these imagers in the lab have in some ways changed very little from the early days. The history of FLIR testing has had its earliest roots from the image intensifier and television community and a tri-service group put together in the early 1970s as a means of developing a standard set of tests that could be used to quantify an IR imager. This paper will chronicle the people and testing techniques used throughout the history of the development of the FLIR from the infancy stages of FLIR testing to the testing methods of today.

The DoD has determined that standardization of Electro-Optic testing is beneficial to current and future Automated Test Systems (ATS). Adopting standards will reduce cost of ownership of ATS and will improve flexibility through interoperability of ATS. The current state of the art in instrument standardization is the Interchangeable Virtual Instrument (IVI) Foundation standards already adopted for commercial standard test equipment such as Digital Multimeters. The Navy has formed a working group entitled “EO Software Working Group” that is meeting quarterly to come up with appropriate IVI standards for EO testing. Considerable progress has been made over the past two years. The first specification, for Blackbodies, has been produced (draft version). Over the next 18 months this will be finalized and submitted to the IVI Standards Committee for formal approval. In addition, other specifications for EO testing will be developed and submitted for formal incorporation. This paper will describe the needs for standardization in the ATS community and the progress in EO IVI standards.

This paper provides an improved method of measuring the modulation transfer function (MTF) in undersampled systems. We show that the currently used canted slit 2D FFT method is limited because interference between the aliased Fourier components and the side-peaks in the non-aliased signal significantly influences the MTF measurements at spatial frequencies larger than the Nyquist frequency. In our new approach, the effective temperature of the slit illumination varies along the slit, with the intensity profile chosen to minimize the interference between the aliased and non-aliased signal components. The accuracy of the measurement procedure has been improved to the point where the main limitation is the temporal and the fixed pattern noise of the IR camera. Experimental results confirming the accuracy at frequencies both below and above the Nyquist frequency are presented.

This paper describes a new test bench for measuring the modulation transfer function of an infrared focal plane array. The system is based on the use of a plane target made with a continuously self-imaging grating (CSIG) that projects in polychromatic light a biperiodic pattern of small and non-diffracting spots called a nondiffracting array.

Minimum Resolvable Temperature Difference (MRTD) is the primary measurement of performance for infrared imaging systems. Where Modulation Transfer Function (MTF) is a measurement of resolution and three-dimensional noise (or noise equivalent temperature difference) is a measurement of sensitivity, MRTD combines both measurements into a test of observer visual acuity through the imager. MRTD has been incorrectly applied to undersampled thermal imagers as a means for assessing the overall performance of the imager. The incorrect application of the MRTD (or just MRT) test to undersampled imagers includes testing to the half-sample (or Nyquist rate) of the sensor and calling the MRT unresolvable beyond this frequency. This approach is known to give poor predictions in overall system performance. Also, measurements at frequencies below the half-sample rate are strongly dependent on the phase between the sampling geometry and the four-bar target. The result is that very little information in the MRT measurement of an undersampled thermal imager is useful. There are a number of alternatives including Dynamic MRT (DMRT), Minimum Temperature Difference Perceived (MTDP), Triangle Orientation Discrimination (TOD), and objective MRT tests. The NVESD approach is to measure the MTF and system noise and to use these measurements in the MRT calculation to give good sensor performance predictions. This paper describes the problems with MRT for undersampled imagers, describes the alternative measurements, and presents the NVESD approach to MRT measurements.

The Minimum Resolvable Temperature Difference (MRT or MRTD) of an IR imaging sensor provides a measure of system performance in terms of sensitivity as a function of resolution. It's expressed as the temperature difference (ΔT) between a target and a background at which target features are just discernable as a function of spatial frequency. Traditionally, MRT has been measured in the laboratory by imaging a flat plate “blackbody” at slightly elevated and depressed temperatures through a sequence of 4-bar slot patterns (“targets”) in high Emissivity disks (“backgrounds”) at ambient temperature. In this traditional, purely emissive MRT approach, the luminance modulation in the images of the smaller targets ride upon luminance pedestals against the ambient background that make the image modulation hard to discern. Consequently, in the laboratory, sensor gain and offset adjustments sometimes must be performed in order to see the modulation on the smaller, i.e., higher spatial frequency, MRT targets. Quite frequently this part of the procedure does not correspond to how the sensor is operated in the field. An alternative approach, called reflective MRT, uses a disk in which the target region consists of slots and highly reflective, low Emissivity spaces and is surrounded by a high Emissivity background. Two flat plate “blackbodies” are used, one in transmission through the slots and one in reflection from the spaces between the slots. Both are at slightly different temperatures controlled and regulated above and below ambient. This results in target luminance modulation that does not ride upon ambient luminance pedestals, thus allowing MRT to be measured at the same sensor gain and offset for all target spatial frequencies. The intent of this approach is to improve the accuracy of laboratory MRT measurements as predictors of field performance. This paper describes this problem with emissive MRT, reflective MRT as a possible solution, and the experimental research planned for calendar year 2003 to compare emissive and reflective MRT measurements of well sampled imaging sensors.

Minimum Resolvable Temperature Difference (MRTD) has long been used to describe the performance of thermal imaging systems. The Visibility Model II developed for second generation thermal imaging systems includes sampling and aliasing issues without assumptions about the observer. As with the earlier reported Visibility Model, applicable for first generation imagers, both objective and subjective measurement schemes can be accommodated. The visibility concept has been demonstrated to be applicable in predicting the objective MRTD. The laboratory measurement of objective MRTD provided the data to evaluate the performance of Objective VISMODII model. Although the measurements are limited in scope, they do demonstrate that the VISMODII predictive model can also be applied to objective MRTD measurements.

The emergence of multi-band sensor technology, e.g. in the thermal infrared, promises significant improvements in TA (target acquisition) performance. With these new sensor systems, targets may be distinguished from their background not only on the basis of differences in radiation magnitude in the sensor's spectral range (as is the case with single-band systems), but also on differences in spectral properties. However, existing end-to-end sensor performance measures, such as the MRTD, MTDP or TOD laboratory tests or the NVTherm model, produce threshold curves of resolution vs. thermal or luminance contrast and do not take spectral difference into account. Until now no test methodology exists to characterize or quantify the additional benefits of a multi-band sensor above a single-band system. We propose an extension to the current end-to-end test methods that may overcome this shortcoming. The method yields a 2-D threshold surface of resolution, contrast and spectral difference between a test pattern and its background. This surface may be used in TA models to predict the ability of a human observer, using the sensor system, to recognnize or identify a target given its size, radiance difference and spectral difference with the background. The extension can be incorporated in the TOD, but in other sensor performance measures as well.

While it is universally recognized that image quality of a thermal sensor is a strong function of spatial uniformity, the metrics commonly used to assess performance do not adequately measure the effectiveness of non-uniformity correction (NUC). Image uniformity is generally not static, particularly if correction terms are updated intermittently (with periodic shuttering) or gradually (with scene-based NUC). Minimum Resolvable Temperature (MRT), the most prevalent test for characterizing overall imaging performance, is poorly suited for characterizing dynamic performance. The Triangle Orientation Discrimination (TOD) metric proposed by Bijl and Valeton, because of its short observation window, provides better capability for evaluating sensors that exhibit non-negligible uniformity drift. This paper compares the effectiveness of MRT and TOD for measuring dynamic performance. TOD measurements of a shutter-based thermal imager are provided immediately after shutter correction and 3 minutes later. The drift in TOD performance shows excellent correlation to drift in system noise.

This paper discusses recent advances in the development of test and evaluation instrumentation for military laser range-finder (LRF) and designation systems. Recent strides have been made at Santa Barbara Infrared (SBIR) in the development of sophisticated active ranging simulation instruments for range accuracy and receiver sensitivity measurement, integrated measurement modules for laser pulse energy and temporal characteristics, and pulsed laser diode targets/sources for shared-aperture IR/laser sensor test and evaluation. In parallel with these activities, NAVSEA has led the development and validation of state-of-the-art reference standard radiometers used in the calibration of narrow-pulse laser systems at 1060 nm and 1550 nm. This paper will describe the application, capabilities, and performance of SBIR's active ranging, laser measurements, and pulsed laser source modules, and NAVSEA's high-performance 1060/1550 nm radiometric instrumentation.

This paper presents qualitative and quantitative comparisons between emissive and reflective target technologies used in the application of IR target projection for thermal imager test and evaluation. Comparison of target projector performance in MRTD, SiTF, MTF, and other test areas will be presented. Relative advantages and disadvantages of emissive and reflective systems will be shown, in addition to requirements placed upon test laboratory environment by the different projector technologies. Discussion of software-based compensation techniques for mitigating reflected ambient effects, environmental ambient drift, and other anomalies will also be provided.

Indigo Systems Corporation has released a commercial off-the-shelf (COTS) PC-based software toolkit named RTools. RTools was developed for engineers and scientists to acquire, radiometrically calibrate, analyze, and document data from most high-end digital infrared (IR) focal plane array (FPA) imaging systems. The RTools software toolkit is comprised of several stand-alone modules including RDac for image acquisition and real-time image analysis, RCal for radiometric and thermographic image calibration, REdit for image file archival and maintenance, and RView for image review, reduction, analysis, and report generation.

Most seeker systems are designed for a single mission, making the cost of the seeker of prime importance. Uncooled detector arrays are significantly less expensive than their cooled counterparts but infrared optics remain a significant cost driver. We have designed and developed an optical system with a long focal length and short physical length that can be fabricated inexpensively; yet deliver the performance required of an infrared seeker. The Folded Molded Mangin is the result of several iterations that yielded a unique optical system that meets those requirements. The system can be fabricated using conventional diamond turning techniques or by molding.

We present preliminary results on the feasibility demonstration of using the rare-earth-doped silica as the IR-to-visible converter. Its principle of operation employs the up-conversion upon pumping with the near IR radiation. The simulation results of the time-dependent 3-D heat transfer study demonstrate the spatial and temporal resolution of this transducer for the constant and impulse irradiation profile, confirming the concept when the Er-doped silica fiber is in contact with a cold reservoir. The preliminary experimental results confirm the feasibility of using the converter for room temperature applications, that correspond to about 10-μm radiation. The error arising from the rare-earth dopant non-uniformity is estimated at +/- 1.5 C.