Review and categorization of spectral processing methods

In this paper numerous known spectral analysis methodologies available to geologic remote sensing are reviewed. According to their requirement for reference data, the techniques are categorized into knowledge-based (KB) or data-driven (DD) approaches. The two categories are comparatively studied and their advantages, disadvantages, and major geologic outcomes are emphasized. Next ways to bridge between the spectral data and geochemical, mineralogic, and lithologic information are noted. Subsequently present and potential hybridization lines among the groups and sub-groups are discussed and at the end the challenges that spectral processing methods should meet are pointed out. In conclusion, hybridization of the two categories is proposed to be one of the lines of attack to yield the most robust processing methodologies, capable of fulfilling the demands of multiand hyperspectral geological remote sensing.