An endmember extraction framework based on abundance constraint

Spectral unmixing is an important technique for hyperspectral data interpretation, in which a mixed spectral signature is decomposed into a collection of spectrally constituent and pure spectra, called endmembers, and a set of correspondent fractions, or abundances, that indicate the proportion of each endmember´s presence in the mixture. As is known to all, we can get abundances with given endmembers. Correspondingly, we can also extract endmembes based on abundance constraints. In this paper, we propose an endmember extraction frame work based on abundance constraints whose efficiency is related to abundance calculation. This new approach has almost the same precision compared with the state-of-art N-FINDR algorithm on both simulated and real data sets with its efficiency better than N-FINDR.