Proc. SPIE. 9820, Infrared Imaging Systems: Design, Analysis, Modeling, and Testing XXVII
KEYWORDS: Targeting Task Performance metric, Signal to noise ratio, Received signal strength, Targeting Task Performance metric, Wavelets, Target acquisition, Signal to noise ratio, Performance modeling, Eye, Image quality, Wavelet transforms, Imaging systems
Target acquisition performance depends strongly on the contrast of the target. The Targeting Task Performance (TTP)
metric, within the Night Vision Integrated Performance Model (NV-IPM), uses a combination of resolution, signal-to-noise
ratio (SNR), and contrast to predict and model system performance. While the dependence on resolution and SNR
are well defined and understood, defining a robust and versatile contrast metric for a wide variety of acquisition tasks is
more difficult. In this correspondence, a wavelet contrast metric (WCM) is developed under the assumption that the
human eye processes spatial differences in a manner similar to a wavelet transform. The amount of perceivable
information, or useful wavelet coefficients, is used to predict the total viewable contrast to the human eye. The WCM is
intended to better match the measured performance of the human vision system for high-contrast, low-contrast, and low-observable
targets. After further validation, the new contrast metric can be incorporated using a modified TTP metric
into the latest Army target acquisition software suite, the NV-IPM.
Measuring the Modulation Transfer Function (MTF) of a display monitor is necessary for many applications such as:
modeling end-to-end systems, conducting perception experiments, and performing targeting tasks in real-word scenarios.
The MTF of a display defines the resolution properties and quantifies how well the spatial frequencies are displayed on a
monitor. Many researchers have developed methods to measure display MTFs using either scanning or imaging devices.
In this paper, we first present methods to measure display MTFs using two separate technologies and then discuss the
impact of a display MTF on a system’s performance. The two measurement technologies were scanning with a photometer
and imaging with a CMOS based camera. To estimate a true display MTF, measurements made with the photometer were
backed out for the scanning optics aperture. The developed methods were applied to measure MTFs of the two types of
monitors, Cathode Ray Tube (CRT) and Liquid Crystal Display (LCD). The accuracy of the measured MTFs was validated
by comparing MTFs measured with the two systems. The methods presented here are simple and can be easily
implemented employing either a Prichard photometer or an imaging device. In addition, the impact of a display MTF on
the end-to-end performance of a system was modeled using NV-IPM.
Measuring the performance of a cathode ray tube (CRT) or liquid crystal display (LCD) is necessary to enable end-to-end system modeling and characterization of currently used high performance analog imaging systems, such as 2nd Generation FLIR systems. If the display is color, the performance measurements are made more difficult because of the underlying structure of the color pixel as compared to a monochrome pixel. Out of the various characteristics of interest, we focus on determining the gamma value of a display. Gamma quantifies the non-linear response between the input gray scale and the displayed luminance. If the displayed image can be corrected for the display’s gamma, an accurate scene can be presented or characterized for laboratory measurements such as MRT (Minimum Resolvable Temperature) and CTF (Contrast Threshold Function). In this paper, we present a method to determine the gamma to characterize a color display using the Prichard 1980A photometer. Gamma corrections were applied to the test images for validating the accuracy of the computed gamma value. The method presented here is a simple one easily implemented employing a Prichard photometer.
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.
Historically, the focus of detection experiments and modeling at the Night Vision and Electronic Sensors Directorate (NVESD) has been on detecting military vehicles in rural terrains. A gap remains in understanding the detection of human targets in rural terrains and how it might differ from detection of vehicles. There are also improvements that can be made in how to quantify the effect of human movement on detectability. Two experiments were developed to look at probability of detection and time to detect fully exposed human targets in a low to moderate clutter environment in the infrared waveband. The first test uses static images of standing humans while the second test uses videos of humans walking across the scene at various ranges and speeds. Various definitions of target and background areas are explored to calculated contrast and target size. Ultimately, task difficulty parameters (V50s) are calculated to calibrate NVESD sensor performance models, specifically NVThermIP and NV-IPM, for the human detection task. The focus of the analysis in this paper is primarily on the static detection task since the analysis for the dynamic detection experiment is still in the early stages. Results will be presented as well as a plan for future work in this area.
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.
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.
To investigate the benefits of multiband infrared sensor design for target detection, search and detection experiments were conducted in the midwave infrared (MWIR) and longwave infrared (LWIR) wavebands in both rural and urban battlefields. In each battlefield environment, real imagery was collected in both bands by a single sensor using the same optics for both bands, resulting in perfect co-registration of the imagery. In order to study the performance impact of the spectral content, and not diffraction or other sensor-specific differences, the images were processed as needed so that differences in resolution due to diffraction were mitigated. The results of perception experiments, including detection probabilities, search times, and false alarm data, were compared between the wavebands.
In an initial effort to better understand how motion in human activities influences sensor performance, Night Vision and Electronic Sensors Directorate (NVESD) developed a perception experiment that tests an observer's ability to identify an activity in static and dynamic scenes. Current sensor models such as NVTherm were calibrated using static imagery of military vehicles but, given the current battlefield environment, the focus has shifted more towards discriminating human activities. In these activities, motion plays an important role but this role is not well quantified by the model. This study looks at twelve hostile and non-hostile activities that may be performed on an urban roadside such as digging a hole, raking, surveillance with binoculars, and holding several weapons. The forced choice experiment presents the activities in both static and dynamic scenes so that the effect of adding motion can be evaluated. The results are analyzed and attempts are made at relating observer performance to various static and dynamic metrics and ultimately developing a calibration for the sensor model.
IR detector integration time is determined by a combination of the scene or target radiance, the noise of the sensor, and the sensor sensitivity. Typical LWIR detectors such as those used in most U.S. military systems can operate effectively with integration times in the microsecond region. MWIR detectors require much longer integration times (up to several milliseconds) under some conditions to achieve good Noise Equivalent Temperature Difference (NETD). Emerging 3rd Generation FLIR systems incorporate both MWIR and LWIR detectors. The category of sensors know as uncooled LWIR require thermal time constants, similar to integration time, in the millisecond range to achieve acceptable good NETD. These longer integration times and time constants would not limit performance in a purely static environment, but target or sensor motion can induce blurring under some circumstances. A variety of tasks and mission scenarios were analyzed to determine the integration time requirements for combinations of sensor platform movement and look angle. These were then compared to the typical integration times for MWIR and LWIR detectors to establish the suitability of each band for the functions considered.
Perception experiments were conducted at Night Vision and Electronic Sensors Directorate (NVESD) to investigate the effect of targets in defilade on the search task. Vehicles were placed in a simulated terrain and were either fully exposed, partially exposed, or placed in hull defilade. These images, along with a number of no-target images, were presented in a time-limited search perception experiment using military observers. The results were analyzed and compared with ACQUIRE predictions to determine if there are factors, other than size, affecting the search task when targets are in defilade.
Recent work by the US Army RDECOM CERDEC Night Vision and Electronic Sensors Directorate (NVESD) has led to the Time-Limited Search (TLS) model, which has given new formulations for the field of view (FOV) search times. The next step in the evaluation of the overall search model (ACQUIRE) is to apply these parameters to the field of regard (FOR) model. Human perception experiments were conducted using synthetic imagery developed at NVESD. The experiments were competitive player-on-player search tests with the intention of imposing realistic time constraints on the observers. FOR detection probabilities, search times, and false alarm data are analyzed and compared to predictions using both the TLS model and ACQUIRE.
The Night Vision and Electronics Sensors Directorate Electro-optics Simulation Toolkit (NVEOST), follow-on to Paint-The-Night, produces real time simulation of IR scenes and sequences using modeled backgrounds and targets with physics and empirically based IR signatures. Range dependant atmospheric effects are incorporated, realistically degrading the infrared scene impinging on an infrared imaging device. Current sensor effects implementation for Paint the Night (PTN) and the Night Vision Image Generator (NVIG) is a 3 step process. First the scene energy is further attenuated by the sensor optic. Second, a prefilter kernel developed off-line, is applied to scenes or frames to affect the sensor modulation transfer function (MTF) "blurring" of scene elements. Thirdly, sensor noise is overlaid on scenes, or more often frames of scenes. NVESD is improving the PTN functionality, now entitled NVEOST, in several ways. In the near future, a sensor effects tool will directly read an NVTHERM input data file, extract that data which it can utilize and then automatically generate the sensor "world view" of a NVEOST scenario. These will include those elements currently employed: optical transmission, parameters used to calculate prefilter MTF (telescope, detector geometry) and temporal-spatial random noise (σTVH). Important improvements will include treatment of sampling effects (under sampling and super-resolution), certain significant postfilters (signal processing including boost and frame integration) and spatial noise. The sensor effects implementation will require minimal interaction; only a well developed NVTHERM input parameter set will be required. The developments described below will enhance NVEOST's utility not only as a virtual simulator but also as a formidable sensor design tool.
In the urban operations (UO) environment, it may be necessary to identify various vehicles that can be referred to as non-traditional vehicles. A police vehicle might require a different response than a civilian vehicle, or a tactical vehicle. This research reports the measured 50% probability of identification cycle criteria (N50s and V50s) required to identify a different vehicle set than previously researched at NVESD. Longwave infrared (LWIR) and midwave infrared (MWIR) imagery of twelve vehicles at twelve different aspects was collected. Some of the vehicles in this confusion set include an ambulance, a police sedan, a HMMWV, and a pickup truck. This set of vehicles represents those commonly found in urban environments. The images were blurred to reduce the number of resolvable cycles. The results of the human perception experiments allowed the 50% probability of identification cycle criteria (N50s and V50s) to be measured. These results will allow the modeling of sensor performance in the urban terrain for infrared imagers.
This research compares target detection in the longwave and midwave spectral bands in urban environments. The Night Vision and Electronic Sensors Directorate (NVESD) imaged one hundred scenes at several Army Military Operations in the Urban Terrain (MOUT) sites during day and night. Images were resized to make the field-of-view (FOV) for each scene approximately the same. These images were then presented in a time-limited search perception experiment using military observers. Probabilities of detection were compared between the two spectral bands. Results from MOUT search were compared with previous modeling efforts.
This paper describes research on the determination of the fifty-percent probability of identification cycle criterion (N50) for two sets of handheld objects. The first set consists of 12 objects which are commonly held in a single hand. The second set consists of 10 objects commonly held in both hands. These sets consist of not only typical civilian handheld objects but also objects that are potentially lethal. A pistol, a cell phone, a rocket propelled grenade (RPG) launcher, and a broom are examples of the objects in these sets. The discrimination of these objects is an inherent part of homeland security, force protection, and also general population security.
Objects were imaged from each set in the visible and mid-wave infrared (MWIR) spectrum. Various levels of blur are then applied to these images. These blurred images were then used in a forced choice perception experiment. Results were analyzed as a function of blur level and target size to give identification probability as a function of resolvable cycles on target. These results are applicable to handheld object target acquisition estimates for visible imaging systems and MWIR systems. This research provides guidance in the design and analysis of electro-optical systems and forward-looking infrared (FLIR) systems for use in homeland security, force protection, and also general population security.