Frame based kernel methods for automatic classification in hyperspectral data

We propose a new kernel and frame based dimension reducing algorithm by exploiting the synergy between endmembers and kernel based classification schemes. This synergy becomes available by means of utilizing the notions of frames and frame representations.