The comparison of pixel size and atmospheric correction method on matched filter detection for a hyperspectral image

Abstract

Two atmospheric correction methods are used to obtain the reflectance for a hyperspectral data image resampled to varying spatial resolution. The physics-based FLAASH approach as well as the in-scene based QUAC method retrieve the reflectance spectra for the scene, and the ability to use the results to detect materials of interest in the image is determined. Using a spectral matched filter to score the results, both FLAASH and QUAC perform well at matching the ground truth spectrum of a bright meter-sized material at ground sampling distances of 2.4–24 m. For a dark material, QUAC performance degrades with lower resolutions.