Expanded dimensionality for image spectroscopy via machine learning

We expand the existing radiance algorithms to a spatiotemporal radiance equation by exploiting multiple hyperspectral image scans acquired by aerial platforms. A machine learning solution is used to learn the relationships between specific surface targets and the parameterization of the radiance equation. The goal is to characterize the atmosphere under different solar and sensor angles, by taking into account pixels for the target as well as nearby pixels. Compared to traditional image spectroscopy, this expanded radiance equation and machine learning solution integrates quantitative mathematical modeling, multiple scanned hyperspectral images and artificial intelligence. The solution is able to model and predict components of the radiance equation with increased spatial and temporal dimensionality as well as improve the target detection in non-ideal conditions, where current solutions based on a single hyperspectral image normally fail. Initial results include an expanded mathematical solution and a discussion of the constraints and assumptions made, as well as results from a sensitivity study performed using synthetic data to test the effect of different vintage points of view on the radiance simulated. Results from the machine learning algorithm show how different combinations of the multiple scans can be used to parameterize the radiance equation and improve on single scan solutions.