Mineral mapping based on independent component analysis for spaceborne hyperspectral data

摘要

Independent component analysis (ICA) model is proposed to extract alteration minerals using spaceborne hyperspectral data, because the result of present methods for alteration minerals identification are affected easily by the factors including excursion and variation of spectral signatures, interference of environment conditions and insufficient spectral library. In our work, the proposed method can realize extraction of alteration minerals under condition that the prior information of mineral spectra is unknown and the background model is not built. Therefore, this method is successfully applied to spaceborne Hyperion data at Qulong district of Tibet, and the application result in our work is approximately in accord with the geological map. Four kinds of minerals have been identified, which include kaolinite, chlorite, rich-aluminium sericite, and poor-aluminium sericite. And then the result that is obtained by ICA model can illuminate the validity and practicability of the proposed method and can provide some useful information and direction for the prognostication of mineral resource.