https://hgpu.org/?p=4772
Parallel implementation of endmember extraction algorithms using NVidia graphical processing units