CIF: Small: Atmospheric Absorption-Based Computational Sensing

Humans and machines make sense of their surroundings first and foremost using vision, and vision is limited when there is a lack of light in the familiar visible spectrum, with wavelengths from 380 nm (violet) to 750 nm (red). However, light is often plentiful in situations when visible light is lacking (such as at night), but most of it is at wavelengths far beyond the red end of the visible spectrum. This long-wave infrared (LWIR) light is responsible for the images produced by common thermal cameras, which are used in search-and-rescue operations, law enforcement, building inspections, and wildlife observation. Those uses are based on sensing only temperature or heat. Broadly, the goal of this project is to do more with LWIR sensing, such as three-dimensional imaging and remote sensing of air composition and temperature. These new capabilities may contribute to improved security, navigation, pollution monitoring, and aeronautical safety. Additional broader impact aspects of the work include workforce training at the high-school, undergraduate, and graduate levels. This project is focused on formulating and solving new inverse problems based on well-studied aspects of radiative transport. The central challenges come from these inverse problems being underdetermined because the degrees of freedom in emissivity, transmittance, and reflected light increase with the number of spectral channels measured. The project will combine physics-based approaches such as parametric modeling of transmittance with learning-based exploitation of scene structure. The specific goals include improving passive ranging and demonstrating the measurement of air temperature and gas concentrations at a distance. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.