Impact of Atmospheric Correction on the Ship Detection Using Airborne Hyperspectral Image

Tae-Sung Kim, Sangwoo Oh, Tae Byung Chun, Moonjin Lee

Abstract: In this study, the effect of hyperspectral atmospheric correction on accuracy of the ship detection was investigated. We applied two atmospheric correction algorithms on airborne hyperspectral data and then detected ship features from reflectance corrected by each algorithms with an unsupervised target detection method. Both ATREM and FLAASH algorithms produce comparable atmospheric correction results but the algorithm which applied with further correction options optimized for coastal monitoring purpose reveals better detection results for ship detection. Compared to the result from ATREM, the result from FLAASH-corrected reflectance shows a distinct contrast between the ship and surrounding background seawater. Mean values of spectral angles between target and endmember spectra were 0.2388 (ATREM) and 0.2169 (FLAASH). This implies a further optimization of the atmospheric correction can improve the performance of target detection using airborne hyperspectral data, especially for maritime search purpose.