Mineral identification in LWIR hyperspectral imagery applying sparse-based clustering

ABSTRACT

An assessment of mineral identification applying hyperspectral infrared imagery in laboratory conditions is presented here and strives to identify nine different minerals (biotite, diopside, epidote, goethite, kyanite, scheelite, smithsonite, tourmaline, quartz). A hyperspectral camera in Long-Wave Infrared (LWIR, 7.7–11.8 $\mu$m) with a LW-macro lens, an infragold plate, and a heating source are instruments used in the experiment. For automated identification, a Sparse Principal Component Analysis (Sparse PCA)-based K-means clustering is employed to categorise all pixel-spectra in different groups. Then the best representatives of each cluster (using spectral averaging) are chosen to compare these spectra to ASTER spectral library of JPL/NASA through spectral comparison techniques. Spectral angle mapper (SAM) and Normalized Cross Correlation (NCC) are two of such techniques, which are used herein to measure the spectral difference. In order to evaluate robustness of clustering results among the minerals spectra, we have added three levels of Gaussian and salt&pepper noise, $0\mathrm{\%},2\mathrm{\%}$, and $4\mathrm{\%}$, to input spectra which dropped the accuracy percentage from more than 84.73$\mathrm{\%}$, for 0$\mathrm{\%}$ added noise, to 44.57$\mathrm{\%}$, for 2$\mathrm{\%}$ average of both additive noise, and 22.21$\mathrm{\%}$, for 4$\mathrm{\%}$ additive noise. The results conclusively indicate a promising performance but noise sensitive behaviour of the proposed approach.

KEYWORDS:

• Hyperspectral imagery
• mineral identification
• sparse principle components analysis
• clustering

Acknowledgements

The authors would thank anonymous reviewers and editors of QIRT journal for their constructive comments. Also we would like to thank Annette SchweThe proposed methodology underlies an application of Sparse PCA in clustering for the segmentation rdtfeger from the Department of Electrical and Computer Engineering at Laval University for her constructive comments and help. We acknowledge Kévin Liaigre and Saeed Sojasi from Laval University for their help in different parts of the project. We also acknowledge Dr. Martin Chamberland vice-president at Telops for his insightful help and collaboration. This research was supported by FRQ-NT (Fonds de Recherche du Québec - Nature et Technologies) grant number: 2014-MI-182452 and conducted under the Canadian tier-1 research chair in Multipolar Infrared Vision (MIVIM), NSERC-Agnico Eagle Industrial Research Chair in Mineral Exploration.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Active thermography is a thermal experiment that involves an external heating sources to preheat (or cooling) specimens [6–8,52,53].

Additional information

Funding

This work was supported by the Laval University [FRQ-NT (Fonds de Recherche du Quebec - Nature et Technologies) grant number: 2014-MI-182452].

Notes on contributors

Bardia Yousefi

Bardia Yousefi received his first Ph.D. in Intelligence Systems from Department of Artificial Intelligence, University of Malaya (UM) at 2016 and his second Ph.D. in Electrical Engineering, Laval University (ULaval) at 2018. He is currently a research faculty at the Center of Biomedical Image Computing and Analytics (CBICA), University of Pennsylvania. He is Vice President of IEEE-Young Professionals (YP) Quebec since 2017 and a member of Computational Breast (Biomedical) Imaging Group (CBIG), Advanced Robotic Laboratory (ARL) at UM, Multipolar Infrared Vision tier 1 Canada Research Chair (MIVIM), Computer Vision and System Laboratory (CVSL) and REPARTI at ULaval, Research Center on Geology and Geological Engineering of Mineral Resources (E4m), NSERC-Agnico Eagle Industrial Research Chair in Mineral Exploration, and US-National Postdoctoral Association (NPA). His main area of interest involves pattern recognition and machine learning. He is a reviewer for more than 20 scientific journals/conferences and has published over 50 research papers.

Clemente Ibarra Castanedo

Clemente Ibarra Castanedo is a professional researcher in the Computer Vision and Systems Laboratory (CVSL) of Laval University in Quebec City, Canada. He received his Ph.D. in Electrical and Computer Engineering from the same institution in 2005. As a member of the Multipolar Infrared Vision Canada Research Chair (MIVIM), he has contributed to several publications in the field of infrared vision. His research interests are in signal processing and image analysis for the nondestructive characterization of materials by active thermography, as well as near and short-wave infrared reflectography/transmittography imaging.

Émilie Bédard

Émilie Bédard holds a M.Sc. degree in Earth Sciences and is a research associate at the Department of Geology and Geological Engineering, Université Laval, Quebec City, QC, Canada. She has been the coordinator of the NSERC—Agnico Eagle Industrial Research Chair in Mineral Exploration as well as of the E4m - Research Center on Geology and Mineral Resources Engineering since 2015. Between 2008 and 2015, she worked as an exploration geologist on different advanced and grassroots projects. She is a registered Geologist with the Québec Order of Geologists.

Georges Beaudoin

Georges Beaudoin was granted BSc and MSc degrees in geology by Université Laval, and a PhD in mineral deposit geochemistry from the University of Ottawa for his study of the silver-lead-zinc veins of the Kokanee Range in BC.  Georges Beaudoin is, since 1993, Professor of Economic Geology at the Département de géologie et de génie géologique of Université Laval. He has led development of the field of magnetite & hematite mineral chemistry and its application to mineral exploration. Since 2012, he holds the NSERC-Agnico Eagle Industrial Research Chair in Mineral Exploration, where his team develops indicator mineral methods for orogenic gold deposits. He is Director of the E4m research Centre on the geology and engineering of mineral resources at Université Laval. Since 2012, he is Editor of Mineralium Deposita and he was President of the Society for Geology Applied to Mineral Deposits, SGA, in 2014-2015.

Xavier P.V. Maldague

Xavier P.V. Maldague, P. Eng. Ph.D., is professor at the Department of Electrical and Computing Engineering of Université Laval, Québec City, Canada. He has trained over 50 graduate students (M.Sc. and Ph.D.) and has more than 300 publications. His research interests are in Infrared Thermography, NonDestructive Evaluation (NDE) techniques and Vision / Digital Systems for industrial inspection. He holds a Tier 1 Canada Research Chair in Infrared Vision. He chairs the Quantitative Infrared Thermography (QIRT) Council. He is an Honorary Fellow of the Indian Society of Nondestructive Testing, a fellow of the Canadian Engineering Institute, American Society of NonDestructive Testing, Alexander von Humbolt Foundation (Germany).