The U.S. Army’s RDECOM CERDEC Night Vision and Electronic Sensors Directorate (NVESD) Perception Lab is tasked with supporting the development of sensor systems for the U.S. Army by evaluating human performance of emerging technologies. Typical research questions involve detection, recognition and identification as a function of range, blur, noise, spectral band, image processing techniques, image characteristics, and human factors. NVESD’s Perception Lab provides an essential bridge between the physics of the imaging systems and the performance of the human operator. In addition to quantifying sensor performance, perception test results can also be used to generate models of human performance and to drive future sensor requirements. The Perception Lab seeks to develop and employ scientifically valid and efficient perception testing procedures within the practical constraints of Army research, including rapid development timelines for critical technologies, unique guidelines for ethical testing of Army personnel, and limited resources. The purpose of this paper is to describe NVESD Perception Lab capabilities, recent methodological improvements designed to align our methodology more closely with scientific best practice, and to discuss goals for future improvements and expanded capabilities. Specifically, we discuss modifying our methodology to improve training, to account for human fatigue, to improve assessments of human performance, and to increase experimental design consultation provided by research psychologists. Ultimately, this paper outlines a template for assessing human perception and overall system performance related to EO/IR imaging systems.
The desire to provide the warfighter both ranging and reflected intensity information is increasing to meet expanding operational needs. LIDAR imaging systems can provide the user with intensity, range, and even velocity information of a scene. The ability to predict the performance of LIDAR systems is critical for the development of future designs without the need to conduct time consuming and costly field studies. Performance modeling of a frequency modulated continuous wave (FMCW) LIDAR system is challenging due to the addition of the chirped laser source and waveform mixing. The FMCW LIDAR model is implemented in the NV-IPM framework using the custom component generation tool. This paper presents an overview of the FMCW Lidar, the customized LIDAR components, and a series of trade studies using the LIDAR model.
Conventional sensors measure the light incident at each pixel in a focal plane array. Compressive sensing (CS) involves capturing a smaller number of unconventional measurements from the scene, and then using a companion process to recover the image. CS has the potential to acquire imagery with equivalent information content to a large format array while using smaller, cheaper, and lower bandwidth components. However, the benefits of CS do not come without compromise. The CS architecture chosen must effectively balance between physical considerations, reconstruction accuracy, and reconstruction speed to meet operational requirements. Performance modeling of CS imagers is challenging due to the complexity and nonlinearity of the system and reconstruction algorithm. To properly assess the value of such systems, it is necessary to fully characterize the image quality, including artifacts and sensitivity to noise. Imagery of a two-handheld object target set was collected using an shortwave infrared single-pixel CS camera for various ranges and number of processed measurements. Human perception experiments were performed to determine the identification performance within the trade space. The performance of the nonlinear CS camera was modeled by mapping the nonlinear degradations to an equivalent linear shift invariant model. Finally, the limitations of CS modeling techniques are discussed.
Conventional electro-optical and infrared (EO/IR) systems capture an image by measuring the light incident at each of
the millions of pixels in a focal plane array. Compressive sensing (CS) involves capturing a smaller number of
unconventional measurements from the scene, and then using a companion process known as sparse reconstruction to
recover the image as if a fully populated array that satisfies the Nyquist criteria was used. Therefore, CS operates under
the assumption that signal acquisition and data compression can be accomplished simultaneously. CS has the potential
to acquire an image with equivalent information content to a large format array while using smaller, cheaper, and lower
bandwidth components. However, the benefits of CS do not come without compromise. The CS architecture chosen
must effectively balance between physical considerations (SWaP-C), reconstruction accuracy, and reconstruction speed
to meet operational requirements.
To properly assess the value of such systems, it is necessary to fully characterize the image quality, including artifacts
and sensitivity to noise. Imagery of the two-handheld object target set at range was collected using a passive SWIR
single-pixel CS camera for various ranges, mirror resolution, and number of processed measurements. Human
perception experiments were performed to determine the identification performance within the trade space. The
performance of the nonlinear CS camera was modeled with the Night Vision Integrated Performance Model (NV-IPM)
by mapping the nonlinear degradations to an equivalent linear shift invariant model. Finally, the limitations of CS
modeling techniques will be discussed.
Atmospheric turbulence is a well-known phenomenon that often degrades image quality due to intensity fluctuations, distortion, and blur in electro-optic and thermal imaging systems. To properly assess the performance of an imaging system over the typical turbulence trade space, a time consuming and costly field study is often required. A fast and realistic turbulence simulation will allow the performance assessment of an imaging system under various turbulence conditions to be done as well as provide input data for the evaluation of turbulence mitigation algorithms in a cost efficient manner. The simulation is based on an empirical model with parameters derived from the first and second-order statistics of imaging distortions measured from field collected data. The dataset consists of image sequences recorded with a variable frame rate visible camera from strong to weak turbulence conditions. The simulation uses pristine, single images containing no turbulence effects as an input and produces image sequences degraded by the specified turbulence. Target range, optics diameter, wavelength, detector integration time, and the wind velocity component perpendicular to the propagation path all contribute to the severity of the atmospheric turbulence distortions and are included in the simulation. The addition of the detector integration time expands the functionality of the simulation tool to include imagers with lower frames rates. Examples are presented demonstrating the utility of the turbulence simulation.
The design and modeling of compressive sensing (CS) imagers is difficult due to the complexity and non-linearity of the system and reconstruction algorithm. The Night Vision Integrated Performance Model (NV-IPM) is a linear imaging system design tool that is very useful for complex system trade studies. The custom component generator, included in NV-IPM, will be used to include a recently published theory for CS that links measurement noise, easily calculated with NV-IPM, to the noise of the reconstructed CS image given the estimated sparsity of the scene and the number of measurements as input. As the sparsity will also depend on other factors such as the optical transfer function and the scene content, an empirical relationship will be developed between the linear model within NV-IPM and the non-linear reconstruction algorithm using measured test data. Using the theory, a CS imager varying the number of measurements will be compared to a notional traditional imager.
The Thermal Range Model (TRM4)1 developed by the Fraunhofer IOSB of Germany is a commonly used performance model for military target acquisition systems. There are many similarities between the U.S Army Night Vision Integrated Performance Model (NV-IPM)2 and TRM4. Almost all of the camera performance characterizations, such as signal-to-noise calculations and modulation transfer theory are identical, only the human vision model and performance metrics differ. Utilizing the new Custom Component Generator in the NV-IPM we develop a component to calculate the Average Modulation at Optimal Phase (AMOP) and Minimum Difference Signal Perceived (MDSP) calculations used in TRM4. The results will be compared with the actual TRM4 results for a variety of imaging systems. This effort demonstrates that the NV-IPM is a versatile system design tool which can easily be extended to include a variety of image quality and performance metrics.
Due to the delay of sequential 3-D Lidar image acquisition while an uncooperative human target is in motion,
the image may generate missing or occlusion pixels. We wish to minimize the impact of image acquisition of
a moving target for aided target recognition. We apply the standard Fourier transform algorithms for an error
resilience restoration to minimize the impact to the Human Visual System (HVS) which tends to overly
emphasize the edge and the artificially generated discontinuity in missing pixels. We compared (i) classical
phase retrieval scheme: Gerchburg-Saxon-Hayes-Papoulis (GSHP) and (ii) the Compressive Sensing scheme:
Candes-Romberg-Donohoe-Tao (CRDT). The following two lessons were learned: The mechanism is based
on Gibbs overshooting of a step-discontinuity. It is based on relocating the sparsely sampled zeros at missing
pixel locations a la spatial and spatial frequency inner product conformal mapping property.
Compressive sensing (CS) can potentially form an image of equivalent quality to a large format, megapixel array, using a smaller number of individual measurements. This has the potential to provide smaller, cheaper, and lower bandwidth imaging systems. To properly assess the value of such systems, it is necessary to fully characterize the image quality, including artifacts, sensitivity to noise, and CS limitations. Full resolution imagery of an eight tracked vehicle target set at range was used as an input for simulated single-pixel CS camera measurements. The CS algorithm then reconstructs images from the simulated single-pixel CS camera for various levels of compression and noise. For comparison, a traditional camera was also simulated setting the number of pixels equal to the number of CS measurements in each case. Human perception experiments were performed to determine the identification performance within the trade space. The performance of the nonlinear CS camera was modeled with the Night Vision Integrated Performance Model (NVIPM) by mapping the nonlinear degradations to an equivalent linear shift invariant model. Finally, the limitations of compressive sensing modeling will be discussed.
The battlefield has shifted from armored vehicles to armed insurgents. Target acquisition (identification, recognition, and detection) range performance involving humans as targets is vital for modern warfare. The acquisition and neutralization of armed insurgents while at the same time minimizing fratricide and civilian casualties is a mounting concern. U.S. Army RDECOM CERDEC NVESD has conducted many experiments involving human targets for infrared and reflective band sensors. The target sets include human activities, hand-held objects, uniforms & armament, and other tactically relevant targets. This paper will define a set of standard task difficulty values for identification and recognition associated with human target acquisition performance.
The standard model used to describe the performance of infrared imagers is the U.S. Army imaging system target
acquisition model, based on the targeting task performance metric. The model is characterized by the resolution and
sensitivity of the sensor as well as the contrast and task difficulty of the target set. The contrast of the target is defined as
a spatial average contrast. The model treats the contrast of the target set as spatially white, or constant, over the bandlimit
of the sensor. Previous experiments have shown that this assumption is valid under normal conditions and typical
target sets. However, outside of these conditions, the treatment of target signature can become the limiting factor
affecting model performance accuracy. This paper examines target signature more carefully. The spatial frequency
dependence of the standard U.S. Army RDECOM CERDEC Night Vision 12 and 8 tracked vehicle target sets is
described. The results of human perception experiments are modeled and evaluated using both frequency dependent and
independent target signature definitions. Finally the function of task difficulty and its relationship to a target set is
Analytical Model 1 describes how long it takes the first observer to find a target when multiple observers search a field of regard using imagery provided
by a single sensor. This model, developed using probability concepts, suggests considerable benefits accrue from collaborative search: when P is near
one and with ten observers the mean detection time (in reduced time) is reduced by almost an order of magnitude when compared to that of a single
observer. To get the instant of detection in clock time we add the delay time td to the reduced time. Empirical fits for td and are also given in the paper.
Model 1 was verified/validated by computer simulation and perception experiments. Here ten observers searched sixty computer generated fields of
regard (each one was 60 x 20 degrees) for a single military vehicle. Analytical Model 2 describes how the probability of target acquisition increases with
the number of observers. The results of Model 2 suggest that probability of target acquisition increases considerably when multiple observers independently
search a field of regard. Model 2 was verified by simulation but not by perception experiment. Models 1 and 2 are pertinent to development of
search strategies with multiple observers and are expected to find use in wargaming for evaluating the efficacy of networked imaging sensors.
The standard model used to describe the performance of infrared sensors is the U.S. Army thermal target acquisition
model, NVThermIP. The model is characterized by the apparent size and contrast of the target, and the resolution and
sensitivity of the sensor. Currently, manual gain and level determine optimal contrast for military targets. The Night
Vision models are calibrated to such images using a spatial average contrast consisting of the root sum squared of the
difference between the target and background means, and the standard deviation of the target internal contrast. This
definition of contrast applied to the model will show an unrealistic increase in performance for saturated targets. This
paper presents a modified definition of target contrast for use in NVThermIP, including a threshold value for target to
background mean difference and means to remove saturated pixels from the standard deviation of the target. Human
perception experiments were performed and the measured results are compared with the predicted performance using the
modified target contrast definition in NVThermIP.
A model has been developed that predicts the probability of detection as a function of time for a sensor on a
moving platform looking for a stationary object. The proposed model takes as input P (calculated from
NVThermIP), expresses it as a function of time using the known sensor-target range and outputs detection
probability as a function of time. The proposed search model has one calibration factor that is determined
from the mean time to detect the target. Simulated imagery was generated that models a vehicle moving
with constant speed along a straight road with varied vegetation on both sides and occasional debris on the
road and on the shoulder. Alongside, and occasionally on the road, triangular and square shapes are visible
with a contrast similar to that of the background but with a different texture. These serve as targets to be
detected. In perception tests, the ability of observers to detect the simulated targets was measured and
excellent agreement was observed between modeled and measured results.
Contrast enhancement and dynamic range compression are currently being used to improve the performance of infrared
imagers by increasing the contrast between the target and the scene content. Automatic contrast enhancement techniques
do not always achieve this improvement. In some cases, the contrast can increase to a level of target saturation. This
paper assesses the range-performance effects of contrast enhancement for target identification as a function of image
saturation. Human perception experiments were performed to determine field performance using contrast enhancement
on the U.S. Army RDECOM CERDEC NVESD standard military eight target set using an un-cooled LWIR camera.
The experiments compare the identification performance of observers viewing contrast enhancement processed images
at various levels of saturation. Contrast enhancement is modeled in the U.S. Army thermal target acquisition model
(NVThermIP) by changing the scene contrast temperature. The model predicts improved performance based on any
improved target contrast, regardless of specific feature saturation or enhancement. The measured results follow the
predicted performance based on the target task difficulty metric used in NVThermIP for the non-saturated cases. The
saturated images reduce the information contained in the target and performance suffers. The model treats the contrast
of the target as uniform over spatial frequency. As the contrast is enhanced, the model assumes that the contrast is
enhanced uniformly over the spatial frequencies. After saturation, the spatial cues that differentiate one tank from
another are located in a limited band of spatial frequencies. A frequency dependent treatment of target contrast is
needed to predict performance of over-processed images.
This paper presents a comparison of the predictions of NVThermIP to human perception experiment results in the
presence of large amounts of noise where the signal to noise ratio is around 1. First, the calculations used in the NVESD
imager performance models that deal with sensor noise are described outlining a few errors that appear in the
NVThermIP code. A perception experiment is designed to test the range performance predictions of NVThermIP with
varying amounts of noise and varying frame rates. NVThermIP is found to overestimate the impact of noise, leading to
pessimistic range performance predictions for noisy systems. The perception experiment results are used to find a best
fit value of the constant α used to relate system noise to eye noise in the NVESD models. The perception results are also
fit to an alternate eye model that handles frame rates below 30Hz and smoothly approaches an accurate prediction of the
performance in the presence of static noise. The predictions using the fit data show significantly less error than the
predictions from the current model.
Contrast enhancement and dynamic range compression are currently being used to improve the performance of infrared
imagers by increasing the contrast between the target and the scene content, by better utilizing the available gray levels
either globally or locally. This paper assesses the range-performance effects of various contrast enhancement algorithms
for target identification with well contrasted vehicles. Human perception experiments were performed to determine field
performance using contrast enhancement on the U.S. Army RDECOM CERDEC NVESD standard military eight target
set using an un-cooled LWIR camera. The experiments compare the identification performance of observers viewing
linearly scaled images and various contrast enhancement processed images. Contrast enhancement is modeled in the US
Army thermal target acquisition model (NVThermIP) by changing the scene contrast temperature. The model predicts
improved performance based on any improved target contrast, regardless of feature saturation or enhancement. To
account for the equivalent blur associated with each contrast enhancement algorithm, an additional effective MTF was
calculated and added to the model. The measured results are compared with the predicted performance based on the
target task difficulty metric used in NVThermIP.
A scanning Fabry-Perot transmission filter composed of a pair of dielectric mirrors has been demonstrated at millimeter
and sub-millimeter wavelengths. The mirrors are formed by alternating quarter-wave optical thicknesses of silicon and
air in the usual Bragg configuration. Detailed theoretical considerations are presented for determining the optimum
design. Characterization was performed at sub-mm wavelengths using a gas laser together with a Golay cell detector and
at mm-wavelengths using a backward wave oscillator and microwave power meter. High resistivity in the silicon layers
was found important for achieving high transmittance and finesse, especially at the longer wavelengths. A finesse value
of 411 for a scanning Fabry-Perot cavity composed of three-period Bragg mirrors was experimentally demonstrated.
Finesse values of several thousand are considered to be within reach. This suggests the possibility of a compact terahertz
Fabry-Perot spectrometer that can operate in low resonance order to realize high free spectral range while simultaneously
achieving a high spectral resolution. Such a device is directly suitable for airborne/satellite and man-portable sensing
With over 110 million landmines buried throughout the world, the ability to detect and identify objects beneath the soil is crucial. The increased use of plastic landmines requires the detection technology to be able to locate both metallic and non-metallic targets. A novel active mmW scanning imaging system was developed for this purpose. It is a hyperspectral system that collects images at different mmW frequencies from 90-140 GHz using a vector network analyzer collecting backscattering mmW radiation from the buried sample. A multivariate statistical method, Principal Components Analysis, is applied to extract useful information from these images. This method is applied to images of different objects and experimental conditions.
Transmission spectra were measured over the range 90-4200 GHz for a locally sourced soil sample composed mostly of quartz sand with ~200 micron particle size. A vector network analyzer covered the spectral range 90-140 GHz. A Fourier spectrometer collected transmission spectra over the range 120 to 4200 GHz. Transmission drops to zero for wavelengths shorter than the characteristic particle size of the sample as a consequence of scattering. Transmission spectra were also measured for various liquids in the 90-140 GHz and 450-1650 GHz ranges in the interest of index matching. These liquids were mixed with the soil sample and were found to reduce scattering and increase transmission through the soil at higher frequencies. This work is relevant to mine detection using THz and millimeter wave (mmW) radiation.
Transmission spectra were measured over the range 90-140 GHz and 300-4200 GHz for 20 soil samples that span a number of soil orders that have extensive worldwide distribution. A vector network analyzer equipped with 16 degree horn antennas covered the spectral range 90-140 GHz. Transmission measurements were also taken for some organic materials in the 90-140 GHz range. A Fourier spectrometer equipped with Hg arc lamp, pellicle beamsplitter, and Si bolometer collected transmission spectra over the range 300 to 4200 GHz. Transmittance ranged from 10-7 to almost 1. In all cases, transmission drops to zero for wavelengths shorter than the characteristic particle size of the sample as a consequence of scattering. In samples of mixed particle size, low transmittance in the 90-140 GHz range was found to be caused by the coarse component. This work is relevant to mine detection using THz and millimeter wave (mmW) radiation.
Multi-layer mirrors capable of >99.9% reflectivity at ~100 micron wavelengths were constructed using thin silicon etalons separated by empty gaps. Due to the large difference between the index of refraction of silicon (3.384) and vacuum (1), calculations indicate that only three periods are required to produce 99.9% reflectivity. The mirror was assembled from high purity silicon wafers, with resistivity over 4000 ohm-cm to reduce free carrier absorption. Wafers were double side polished with faces parallel within 10 arc seconds. The multi-layer mirror was demonstrated as a cavity mirror for the far-infrared p-Ge laser.
A far-infrared p-type germanium laser with active crystal prepared from ultra pure single-crystal Ge by neutron transmutation doping (NTD) is demonstrated. Calculations show that the high uniformity of Ga acceptor distribution achieved by NTD significantly improves average gain. The negative factor of stronger ionized impurity scattering due to high compensation in NTD Ge is shown to be unremarkable for the gain at moderate doping concentrations sufficient for laser operation. Experimentally, this first NTD laser is found to have lower current-density lasing threshold than the best of a number of melt-doped laser crystals studied for comparison.
An etched silicon gold plated lamellar mirror is demonstrated as a fixed-wavelength intracavity selector for the far-infrared p-Ge laser, facilitating spectroscopic applications. The depth of the selective mirror, which defines the laser operation wavelength, can be precisely controlled during the etching process. The third-order Fabry-Perot resonance of this selector yields an active cavity finesse of at least 0.06.
A thin two-side polished silicon etalon is demonstrated as a fixed-wavelength intracavity selector for the far-infrared p-Ge laser. The active cavity finesse is ~ 0.1. The wavelength position and spectral purity are maintained over a wide range of laser operating fields. A p-Ge laser with such a selector may find application in chemical sensing, THz imaging, or non-destructive testing.