The long-wave infrared (LWIR) hyperpectral sensing modality is one that is often used for the problem
of detection and identification of chemical warfare agents (CWA) which apply to both military
and civilian situations. The inherent nature and complexity of background clutter dictates a need
for sophisticated and robust statistical models which are then used in the design of optimum signal
processing algorithms that then provide the best exploitation of hyperspectral data to ultimately make
decisions on the absence or presence of potentially harmful CWAs. This paper describes the basic
elements of an automated signal processing pipeline developed at MIT Lincoln Laboratory. In addition
to describing this signal processing architecture in detail, we briefly describe the key signal models
that form the foundation of these algorithms as well as some spatial processing techniques used for
false alarm mitigation. Finally, we apply this processing pipeline to real data measured by the Telops
FIRST hyperspectral (FIRST) sensor to demonstrate its practical utility for the user community.
Detection performance of LWIR passive standoff chemical agent sensors is strongly influenced by various scene parameters, such as atmospheric conditions, temperature contrast, concentration-path length product (CL), agent absorption coefficient, and scene spectral variability. Although temperature contrast, CL, and agent absorption coefficient affect the detected signal in a predictable manner, fluctuations in background scene spectral radiance have less intuitive consequences. The spectral nature of the scene is not problematic in and of itself; instead it is spatial and temporal fluctuations in the scene spectral radiance that cannot be entirely corrected for with data processing. In addition, the consequence of such variability is a function of the spectral signature of the agent that is being detected and is thus different for each agent. To bracket the performance of background-limited (low sensor NEDN), passive standoff chemical sensors in the range of relevant conditions, assessment of real scene data is necessary1. Currently, such data is not widely available2. To begin to span the range of relevant scene conditions, we have acquired high fidelity scene spectral radiance measurements with a Telops FTIR imaging spectrometer3. We have acquired data in a variety of indoor and outdoor locations at different times of day and year. Some locations include indoor office environments, airports, urban and suburban scenes, waterways, and forest. We report agent-dependent clutter measurements for three of these backgrounds.
In order to meet current and emerging needs for remote passive standoff detection of chemical agent threats, MIT Lincoln Laboratory has developed a Wide Area Chemical Sensor (WACS) testbed. A design study helped define the initial concept, guided by current standoff sensor mission requirements. Several variants of this initial design have since been proposed to target other applications within the defense community. The design relies on several enabling technologies required for successful implementation. The primary spectral component is a Wedged Interferometric Spectrometer (WIS) capable of imaging in the LWIR with spectral resolutions as narrow as 4 cm-1. A novel scanning optic will enhance the ability of this sensor to scan over large areas of concern with a compact, rugged design. In this paper, we shall discuss our design, development, and calibration process for this system as well as recent testbed measurements that validate the sensor concept.
We consider the problem of remotely identifying gaseous materials using passive sensing of long-wave infrared (LWIR) spectral features at hyperspectral resolution. Gaseous materials are distinguishable in the LWIR because of their unique spectral fingerprints. A sensor degraded in capability by noise or limited spectral resolution, however, may be unable to positively identify contaminants, especially if they are present in low concentrations or if the spectral library used for comparisons includes materials with similar spectral signatures. This paper will quantify the relative importance of these parameters and express the relationships between them in a functional form which can be used as a rule of thumb in sensor design or in assessing sensor capability for a specific task.
This paper describes the simulation of remote sensing datacontaining a gas cloud.In each simulation, the spectra are degraded in spectral resolution and through the addition of noise to simulate spectra collected by sensors of varying design and capability. We form a trade space by systematically varying the number of sensor spectral channels and signal-to-noise ratio over a range of values. For each scenario, we evaluate the capability of the sensor for gas identification by computing the ratio of the F-statistic for the truth gas tothe same statistic computed over the rest of the library.The effect of the scope of the library is investigated as well, by computing statistics on the variability of the identification capability as the library composition is varied randomly.