Christopher M. Gittins

Tunable Fabry–Perot etalon-based long-wavelength infrared imaging spectroradiometer

Imaging spectrometry enables passive, stand-off detection and analysis of the chemical composition of gas plumes and surfaces over wide geographic areas. We describe the use of a long-wavelength infrared imaging spectroradiometer, comprised of a low-order tunable Fabry-Perot etalon coupled to a HgCdTe detector array, to perform multispectral detection of chemical vapor plumes. The tunable Fabry-Perot etalon used in this research provides coverage of the 9.5-14-microm spectral region with a resolution of 7-9 cm(-1). The etalon-based imaging system provides the opportunity to image a scene at only those wavelengths needed for chemical species identification and quantification and thereby minimize the data volume necessary for selective species detection. We present initial results using a brassboard imaging system for stand-off detection and quantification of chemical vapor plumes against near-ambient-temperature backgrounds. These data show detection limits of 22 parts per million by volume times meter (ppmv x m) and 0.6 ppmv x m for dimethyl methyphosphonate and SF6, respectively, for a gas/background DeltaT of 6 K. The system noise-equivalent spectral radiance is approximately 2 microW cm(-2) sr(-1) microm(-1). Model calculations are presented comparing the measured sensitivity of the sensor to the anticipated signal levels for two chemical release scenarios.

Band selection in hyperspectral imagery using sparse support vector machines

In this paper we propose an ι1-norm penalized sparse support vector machine (SSVM) as an embedded approach to the hyperspectral imagery band selection problem. SSVMs exhibit a model structure that includes a clearly identifiable gap between zero and non-zero weights that permits important bands to be definitively selected in conjunction with the classification problem. The SSVM Algorithm is trained using bootstrap aggregating to obtain a sample of SSVM models to reduce variability in the band selection process. This preliminary sample approach for band selection is followed by a secondary band selection which involves retraining the SSVM to further reduce the set of bands retained. We propose and compare three adaptations of the SSVM band selection algorithm for the multiclass problem. Two extensions of the SSVM Algorithm are based on pairwise band selection between classes. Their performance is validated by using one-against-one (OAO) SSVMs. The third proposed method is a combination of the filter band selection method WaLuMI in sequence with the (OAO) SSVM embedded band selection algorithm. We illustrate the perfomance of these methods on the AVIRIS Indian Pines data set and compare the results to other techniques in the literature. Additionally we illustrate the SSVM Algorithm on the Long-Wavelength Infrared (LWIR) data set consisting of hyperspectral videos of chemical plumes.

Detection and characterization of chemical vapor fugitive emissions by nonlinear optimal estimation: theory and simulation

This paper addresses detection and characterization of chemical vapor fugitive emissions in a nonscattering atmosphere by processing of remotely-sensed long-wavelength infrared spectra. The analysis approach integrates a parameterized signal model based on the radiative transfer equation with a statistical model for the infrared background. The maximum likelihood model parameter values are defined as those that maximize a Bayesian posterior probability and are estimated using a Gauss-Newton algorithm. For algorithm performance evaluation we simulate observation of fugitive emissions by augmenting plume-free measured spectra with synthetic plume signatures. As plumes become optically thick, the Gauss-Newton algorithm yields significantly more accurate estimates of chemical vapor column density and significantly more favorable plume detection statistics than clutter-matched-filter-based and adaptive-subspace-detector-based plume characterization and detection.

LWIR multispectral imaging chemical sensor

Physical Sciences Inc. (PSI) has developed an Adaptive IR Imaging Spectroradiometer, comprised of a low-order tunable Fabry-Perot etalon coupled to an HgCdTe detector array, for passive, stand-off detection of chemical vapor plumes. The tunable etalon allows coverage of the 9.5 to 14 micrometers spectral region with a resolution of approximately 7 cm-1 and provides the capability to obtain monochromatic images of a scene at only those wavelengths needed for chemical species identification and quantification. The adaptive sampling capability of the etalon allows suppression of background clutter and minimization of data volume. The tuning time between transmission wavelengths is typically approximately 10 ms, however the mirror tuning system may be operated to obtain tuning times as short as 1.3 ms. We present results using a brassboard imaging system for stand-off detection and visualization of chemical vapor plumes against near ambient temperature backgrounds. This data shows detections limits of 22 ppmv m and 0.6 ppmv m for DMMP and SF6 respectively against a (Delta) T of 6 K. The reported detection limits are consistent with the measured system noise-equivalent spectral radiance, approximately 2 (mu) W cm-2 sr-1 micrometers -1.

Passive and active standoff infrared detection of bio‐aerosols

Biological compounds are known to have infrared spectra indicative of specific functional groups. There is a strong interest in the use of passive means to detect airborne biological particles, such as spores and cells, which may act as biological weapons. At the sizes of interest, the infrared spectra of bacterial particles result from a combination of geometric (πdparticle > λ) and Mie (πdparticle ∼ λ) scattering processes, whereas the infrared spectrum of atmospheric particles falls into the Rayleigh limit (πdparticle ≪ λ). In this article we report on laboratory measurements of the infrared spectra of aerosolized Bacillus subtilis (BG) spores in air under controlled measurement conditions. Transmission measurements show an IR spectrum of the spores with features comparable to the condensed phase spectrum superimposed on a background of Mie scattering. Preliminary measurements indicate a peak extinction coefficient of approximately 1.6 × 10−8 cm2 per spore at 9.65 μm. These results are discussed in terms of their implication for passive and active infrared detection and identification of bio-aerosols.

Remote sensing and selective detection of chemical vapor plumes by LWIR imaging Fabry-Perot spectrometry

Physical Sciences Inc. has developed and tested two long-wavelength infrared (LWIR) hyperspectral imaging spectroradiometers based on the insertion of a rapidly tunable Fabry-Perot etalon in the field of view of a HgCdTe focal plane array (FPA). The tunable etalon-based optical system enables a wide field-of-view and the acquisition of narrowband (7 to 11 cm-1 spectral resolution), radiometrically calibrated imagery throughout the 8 to 11 micrometers spectral region. The instruments function as chemical imaging sensors by comparing the spectrum of each pixel in the scene with reference spectra of target chemical species. We present results of recent field tests in this paper.

Imaging sensor constellation for tomographic chemical cloud mapping.

A sensor constellation capable of determining the location and detailed concentration distribution of chemical warfare agent simulant clouds has been developed and demonstrated on government test ranges. The constellation is based on the use of standoff passive multispectral infrared imaging sensors to make column density measurements through the chemical cloud from two or more locations around its periphery. A computed tomography inversion method is employed to produce a 3D concentration profile of the cloud from the 2D line density measurements. We discuss the theoretical basis of the approach and present results of recent field experiments where controlled releases of chemical warfare agent simulants were simultaneously viewed by three chemical imaging sensors. Systematic investigations of the algorithm using synthetic data indicate that for complex functions, 3D reconstruction errors are less than 20% even in the case of a limited three-sensor measurement network. Field data results demonstrate the capability of the constellation to determine 3D concentration profiles that account for ~?86%? of the total known mass of material released.

AIRIS standoff multispectral sensor

The AIRIS Wide Area Detector is an imaging multispectral sensor that has been successfully tested in both ground and airborne configurations for the detection of chemical and biological agent simulants. The sensor is based on the use of a Fabry-Perot based tunable filter with a 256x256 pixel HgCdTe focal plane array providing a 32x32 degree field of regard with 10 meter spatial resolution at a range of 5 km. The sensor includes a real-time processor that produces an infrared image of the scene under interrogation overlaid with color-coded pixels indicating the identity and location of simulants detected by the sensor. We review test data from this sensor taken at Dstl Porton Down, NSWC Dahlgren, as well as from multiple test entries at Dugway Proving Ground. The data indicate the ability to detect release quantities from 0.15 to 360 kg at ranges of ~ 4.7 km including simultaneous multi-simulant releases.

Improved detection and false alarm rejection for chemical vapors using passive hyperspectral imaging

Two AIRIS sensors were tested at Dugway Proving Grounds against chemical agent vapor simulants. The primary objectives of the test were to: 1) assess performance of algorithm improvements designed to reduce false alarm rates with a special emphasis on solar effects, and 3) evaluate performance in target detection at 5 km. The tests included 66 total releases comprising alternating 120 kg glacial acetic acid (GAA) and 60 kg triethyl phosphate (TEP) events. The AIRIS sensors had common algorithms, detection thresholds, and sensor parameters. The sensors used the target set defined for the Joint Service Lightweight Chemical Agent Detector (JSLSCAD) with TEP substituted for GA and GAA substituted for VX. They were exercised at two sites located at either 3 km or 5 km from the release point. Data from the tests will be presented showing that: 1) excellent detection capability was obtained at both ranges with significantly shorter alarm times at 5 km, 2) inter-sensor comparison revealed very comparable performance, 3) false alarm rates < 1 incident per 10 hours running time over 143 hours of sensor operations were achieved, 4) algorithm improvements eliminated both solar and cloud false alarms. The algorithms enabling the improved false alarm rejection will be discussed. The sensor technology has recently been extended to address the problem of detection of liquid and solid chemical agents and toxic industrial chemical on surfaces. The phenomenology and applicability of passive infrared hyperspectral imaging to this problem will be discussed and demonstrated.

AIRIS multispectral imaging chemical sensor

The adaptive IR imaging spectroradiometer (AIRIS) is a multispectral imaging system comprising a low-order tunable Fabry-Perot etalon coupled to an IR focal plane array. This low-order interferometer based imaging system provides wide spectral coverage combined with narrow spectral bandwidth, flexible and adaptive sampling and processing of the image to isolate specific spectral features or signatures, high radiance sensitivity, and an extended field-of-view for the survey of wide areas. The adaptive sampling capability of the AIRIS sensor provides the opportunity to rapidly image a scene at only those wavelengths needed for target identification and clutter suppression. We have developed a prototype LWIR AIRIS sensor to perform passive stand-off detection of hazardous chemical vapor plumes. The imaging sensor covers the 10.0 to 11.5 micrometers region and allows identification of numerous compounds, including chemical warfare agents and simulants, on the basis of observed IR spectra. The LWIR AIRIS has a 40 X 40 deg FOV and a NESR equals 2 (mu) W cm-2 sr-1, resulting in a detection limit of 25 ppmv*m for DMMP against a temperature drop of 6 degrees C.