Description:
OBJECTIVE: Demonstrate and deliver a novel, noncontacting, broad area rapid scanning surface contamination sensor to provide threat warning in real time. DESCRIPTION: The LWIR (long wave infrared) portion of the spectrum possesses absorption, backscatter, and radiation features that can be used with some limited success to detect and identify chemical agents on surfaces. Passive hyperspectral imaging at LWIR is one such approach that uses the sun as the primary illuminator; however, the signal-to-noise ratio (SNR) per pixel is relatively low. On the other hand, a wavelength-agile LWIR laser could provide high SNR per pixel. One possible laser source is the CO2 type which can provide wavelength diversity over approximately 60 lines in the 9.3-10.7 micron band. A laser source of this kind may need to be expanded in its wavelength diversity, possibly by use of isotopic mixtures to achieve perhaps 120 lines, approximating a high intensity continuous broadband illuminator. The laser would have to be made highly divergent in order to illuminate the scene in a single shot. A compact CO2 laser can probably be made powerful enough to cover a relatively large field of view at high intensity for multiple wavelength (active hyperspectral) imaging. It is likely that fusion of data from an intense laser source with dense wavelength diversity (approaching a continuous source such as the sun) and a passive hyperspectral channel would provide significantly enhanced SNR compared to either channel alone. It may be possible that the laser source could provide the required SNR alone. Advanced detection algorithms would be needed to maximize the likelihood of detection in a single channel, and then in the combined fusion channel. It will be essential that the algorithms operate in real time, suggesting that they be robust and not overly complex. Recent advances in fast algorithms suggest that this will be possible. In the case of a combined active and passive sensor, a complication in detecting surface contamination is the presence of the natural background reflection (active) and emission (LWIR passive) from the surface itself. Because of the small signals at contamination levels that must be detected, a detection algorithm using a physics-based radiative transport (RT) model that includes both emission and reflection components from the natural background and contamination would be useful. The multispectral passive image component of the RT model could lead to a method for estimating both spectral and spatial components for two or more background materials. An adaptation of the methods used for unmixing multiple aerosols from multiple wavelength lidar backscatter data could be useful for the background estimation task. A Kalman filtering approach would provide fast real time processing. Previously developed models for chemical detection with FLIRs may be adaptable to surface reflection and emission for the complete active/passive sensor configuration. Recent field testing with the ECBC FAL (Frequency Agile Laser) sensor and CO2 TEA (Transversely Excited Atmospheric) laser transmitter has demonstrated that airborne chemical vapors, chemical aerosols, and biological particles can be simultaneously detected. If surface contaminant detection at LWIR proves feasible, then it will be possible to detect a large number of agents in various forms with a single sensor. Passive hyperspectral sensors, cannot detect biological particles or chemical aerosols. PHASE I: Develop a physics-based model for the combined active/passive sensor under background-only and background plus contamination cases. Using that model, develop an algorithm for estimating the background signal components and develop a prototype detection algorithm that could be generalized in Phase II to a real-time processor. Develop performance and sensor design analysis for a laboratory proof-of-principle surface sensor. Assemble the demonstrator and operate it under direct sunlight to obtain a data base for surface detection at a range not less than 1 m and for at least a two component mixture of interferent and agent simulant. Apply the proof-of-principle algorithms to the data base and demonstrate surface detection with discrimination between the (at least) two mixture components. Based on the initial sensor performance, assemble a detailed roadmap for further development of algorithms and critical sensor components for a compact, real time sensor brassboard to be fabricated, demonstrated, and delivered in the Phase II program. PHASE II: Use the results of the Phase I effort as a data base to develop detection algorithms that operate in real time and to develop a fieldable brassboard sensor. Fabricate the brassboard sensor and demonstrate that it meets the proposed detection sensitivity goals. Demonstrate the sensor in an external ambient environment under direct sunlight. Deliver the sensor in brassboard form suitable for field testing by the Army. Provide advanced sensor concept analysis to show that a prototype sensor can be built with a volume of not greater than 4 cu. ft. and capable of real-time detection/identification of multiple agents on surfaces. PHASE III: The novel Advanced Surface Contamination Sensor delivered under this program would be put into field trials to develop a performance data base for all agent targets. The data base would support parallel development of advanced algorithms for simultaneous detection/identification of the specific agent types and for development of integrated sensor operational protocols. The combination of sensor performance data, advanced algorithms, and sensor operation procedures would form the basis for development of a preproduction Advanced Engineering Model for military deployment, civilian homeland defense, and environmental monitoring. Field demonstration of the advanced prototype(s) would provide the basic information for formulation of a development consortium including private industry and the government. PHASE III DUAL-USE APPLICATIONS: The Advanced Surface Contamination Sensor would fill important roles in rapid threat detection with a noncontact, compact sensor for homeland security and environmental monitoring for which there are presently no adequate solutions. REFERENCES: 1. R. Warren, S. Osher, and R. Vanderbeek,"Multiple aerosol unmixing using the split Bregman algorithm", to appear in Trans. Geoscience and Remote Sensing. 2. M. Althouse and C. Cheng,"Chemical vapor detection with a multispectral thermal imager", Optical Engineering, vol. 30, no. 11, pp. 1725-1733 (1991). 3. R. Warren, R. Vanderbeek, and J. Ahl,"Online estimation of vapor path-integrated concentration and absorptivity using multi-wavelength differential absorption lidar", Applied Optics, Vol. 46, No. 31, pp 7579-7586, 2007. 4. R. Warren, R. Vanderbeek, A. Ben-David, and J. Ahl"Simultaneous estimation of aerosol cloud concentration and spectral backscatter from multiple-wavelength lidar data", Applied optics, Vol. 47, No. 24, pp 4309-4320, 2008 5. Warren, R. Vanderbeek, and J. Ahl,"Estimation and discrimination of aerosols using multiple-wavelength LWIR lidar,"R. SPIE Conference 7665, Chemical, Biological, Radiological, Nuclear, and Explosives Sensing XI, Orlando, FL, April 2010 6. R. Vanderbeek, R. Warren, and J. Ahl,"LWIR Differential Scattering Discrimination of Bio-Aerosols"Seventh Joint Conference on Standoff Detection for Chemical and Biological Defense, Williamsburg, VA, Oct 23-27 (2006). 7. D. Cohn, J. Fox, and C. Swim,"Frequency agile CO2 laser and chemical sensor", IRIS Active Sensor Conference, Monterey, CA, Nov. 1993.
