Performs library pruning based hyperspectral unmixing using recursive PCA %Library Pruning by Recursive PCA
This method identifies which spectral library elements are image endmembers. It performs spectral library pruning.
In this work, Data: HSI data of size NoPix,NoBand Lib: Spectral library to be pruned (of size NoLibEle,NoBand) NoEm: Number of spectral library elements to be pruned
FIrst estimate the number of endmembers (NoEm) by methods such as- HFC-VD [1], GapVD [2] etc. After pruning the library, run SUnSAL algorithm to estiamte the abundance of the pruned library endmembers.
[1] Estimation of number of spectrally distinct signal sources in hyperspectral imagery [2] Noise robust estimation of number of endmembers in a hyperspectral image by eigenvalue based gap index