Abstract: Infrared imaging spectrometers are frequently used for detecting chemicals at standoff distances. Cost, size, and sensitivity are common tradeoffs in this regime, particularly when deploying infrared imaging arrays. In this work, we develop and characterize an infrared snapshot computational imaging spectrometer that leverages a multi-aperture filtered design. A theoretical model is developed, describing the multiplexed encoding technique. The experimental system is then described, including filter optimization and fabrication. Finally, the performance of the system is tested, leveraging a neural-network-based calibration approach, for various indoor and outdoor detection scenarios involving liquid contaminants. The results of our testing demonstrate that the system can detect room-temperature liquid contaminants under cold sky downwelling radiance conditions. We achieve a false positive rate (FPR) of 0.12% at a true positive rate (TPR) of 95% for silicon oil on sand at 18°C and a FPR of 2% at a TPR of 95% for silicon oil on various substrates at 23°C. Results support the efficacy of using uncooled polymer absorption filters for infrared imaging liquid contaminant detectors.
B. D. Maione, C. Baldridge, and M. W. Kudenov, “Microbolometer with a multi-aperture polymer thin-film array for neural-network-based target identification,” Appl. Opt., AO 58(27), 7285–7297 (2019).