Hyperspectral compressed sensing using for endmember extraction

Hyperspectral imagery processing typically demands enormous computational resources in terms of storage, computation, and I/O throughputs, due to its huge amount of data. A special kind of sensing matrices is constructed for compressed sensing imaging of hyperspectral imagery. Endmember can be extracted directly from the compressed hyperspectral data using the proposed sensing matrices with no need to reconstruct full hyperspectral images. Synthetic and real hyperspectral data experiments show that the proposed method has a high potential in endmember extraction from compressed hyperspectral data.