Autonomous Atmospheric Correction Algorithm for Long Wave Infrard Hyperspectral Imagery

摘要

This paper proposes an approach to atmospheric correction for hyperspectral imagery in the long wave infrared spectral band Atmospheric correction requires the estimation of the atmospheric optical parameters (AOP): the transmittance, the path radiance and the downwelling irradiance. Then a temperature emissivity separation process is used to extract the temperature and the emissivity (TES) of the imaged pixels. To estimate the AOP three measurements are required which provide contrast in temperature and emissivity. These measurements need to be as noise free as possible. As such, averaging is required and pixels of similar nature need to be grouped The Radiance Classification Autonomous Atmospheric Correction (RACAAC) algorithm described below first operates a classification on the pixels to obtain two groups of pixels that are close to blackbodies. It then detects pixels that are reflective in nature. With these groups of pixels, it estimates the AOPs and then performs a TES operation on the image.