Microbolometer with a multi-aperture polymer thin-film array for neural-network-basedtarget identification.

Area of Science:

• Optics and Photonics
• Spectroscopy
• Computational Imaging

Background:

• Infrared imaging spectrometers are crucial for standoff chemical detection.
• Existing systems face tradeoffs in cost, size, and sensitivity.
• Developing compact, sensitive, and cost-effective solutions is essential.

Purpose of the Study:

• To develop and characterize an infrared snapshot computational imaging spectrometer.
• To address limitations of traditional infrared imaging arrays.
• To enable effective standoff detection of liquid contaminants.

Main Methods:

• Utilized a multi-aperture filtered design for multiplexed encoding.
• Developed a theoretical model for the imaging technique.
• Optimized and fabricated uncooled polymer absorption filters.
• Employed a neural-network-based calibration approach for performance testing.

Main Results:

• Demonstrated detection of room-temperature liquid contaminants under challenging conditions (e.g., cold sky downwelling radiance).
• Achieved a 0.12% false positive rate (FPR) at 95% true positive rate (TPR) for silicon oil on sand at 18°C.
• Achieved a 2% FPR at 95% TPR for silicon oil on various substrates at 23°C.

Conclusions:

• The developed infrared snapshot computational imaging spectrometer is effective for liquid contaminant detection.
• Uncooled polymer absorption filters are viable for infrared imaging liquid contaminant detectors.
• The system shows promise for standoff chemical detection in diverse environments.