A spatial/spectral protocol for quality assurance of decompressed hyperspectral data for practical applications

Abstract

A detailed spatial/spectral protocol to test the influence of compressed-decompressed on the hyperspectral image data is presented. The proposed scheme is evaluated by applying an hybrid algorithm (LCT and SPHIT) on AISA-Dual hyperspectral images. For the purpose of reliable thematic results, the preservation and recovering of reflectance spectral features should be on the main concern. Therefore, this protocol contains three stages of data processing: radiance, radiometric/spectral preprocessed radiance, and atmospheric corrected reflectance (generated from radiometric/spectral preprocessed radiance). As for impact on exploitation, we consider IsoData classification, anomaly-detection, and spectral indices as benchmark applications. Additionally, the proposed scheme is compared with common validation approaches, such as spectral response assessment and anomaly-detection techniques.