Unsupervised Identification of Targeted Spectra Applying Rank1-NMF and FCC Algorithms in Long-Wave Hyperspectral Infrared Imagery

Clustering methods unequivocally show considerable influence on many recent algorithms and play an important role in hyperspectral data analysis. Here, we challenge the clustering for mineral identification using two different strategies long wave infrared (LWIR, 7.7–11.8 ?m). For that, compare to perform a unique dataset. The first algorithm uses spectral comparison techniques all pixel-spectra creates RGB false color composites (FCC). Then, based is used group regions (called FCC-clustering). second clusters directly spectra. rank of non-negative matrix factorization (NMF) extracts representative each cluster compares results with library JPL/NASA. These give values as features which convert into RGB-FCC rank1-NMF). We applied K-means approach, can be modified any other similar approach. clustering-rank1-NMF indicate significant computational efficiency (more than 20 times faster previous approach) promising performance having up 75.8% 84.8% average accuracies FCC-clustering clustering-rank1 NMF (using angle mapper (SAM)), respectively. Furthermore, several are also such adaptive matched subspace detector (AMSD), orthogonal projection (OSP) algorithm, principal component analysis (PCA), local filter (PLMF), SAM, normalized cross correlation (NCC) both most them range accuracy. However, SAM NCC preferred due their simplicity. Our strive identify eleven grains (biotite, diopside, epidote, goethite, kyanite, scheelite, smithsonite, tourmaline, pyrope, olivine, quartz).