EGU2020-19667, updated on 12 Jun 2020
https://doi.org/10.5194/egusphere-egu2020-19667
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Mineral interpretation results using deep learning with hyperspectral imagery

Andrés Bell1, Carlos Roberto Del-Blanco1, Fernando Jaureguizar1, Narciso García1, and María José Jurado2
Andrés Bell et al.
  • 1ETSI Telecomunicación - Señales, Sistemas y Radiocomunicaciones, Universidad Politécnica de Madrid, Madrid, Spain
  • 2Institute of Earth Sciences Jaume Almera, CSIC - Geophysics & Geohazards, Barcelona, Spain

Minerals are key resources for several industries, such as the manufacturing of high-performance components and the latest electronic devices. For the purpose of finding new mineral deposits, mineral interpretation is a task of great relevance in mining and metallurgy sectors. However, it is usually a long, costly, laborious, and manual procedure. It involves the characterization of mineral samples in laboratories far from the mineral deposits and it is subject to human interpretation mistakes. To address the previous problems, an automatic mineral recognition system is proposed that analyzes in real-time hyperspectral imagery acquired in different spectral ranges: VN-SWIR (Visible, Near and Short Wave Infrared) and LWIR (Long Wave Infrared). Thus, more efficient, faster, and more economic explorations are performed, by analyzing in-situ mineral deposits in the subsurface, instead of in laboratories. The developed system is based on a deep learning technique that implements a semantic segmentation neural network that considers spatial and spectral correlations. Two different databases composed by scanned drilled mineral cores from different mineral deposits have been used to evaluate the mineral interpretation capability. The first database contains hyperspectral images in the VN-SWIR range and the second one in the LWIR range. The obtained results show that the mineral recognition for the first database (VN-SWIR band) achieves an 86% in accuracy considering the following mineral classes: Actinolite, amphibole, biotite-chlorite, carbonate, epidote, saponite, whitemica and whitemica-chlorite. For the second database (LWIR band), a 90% in accuracy has been obtained with the following mineral classes: Albite, amphibole, apatite, carbonate, clinopyroxene, epidote, microcline, quartz, quartz-clay-feldspar and sulphide-oxide. The mineral recognition capability has been also compared between both spectral bands considering the common minerals in both databases. The results show a higher recognition performance in the LWIR band, achieving a 96% in accuracy, than in the VN-SWIR bands, which achieves an accuracy of 85%. However, the hyperspectral cameras covering VN-SWIR range are significantly more economic than those covering the LWIR range, and therefore making them a very interesting option for low-budget systems, but still with a good mineral recognition performance. On the other hand, there is a better recognition capability for those mineral categories with a higher number of samples in the databases, as expected. Acknowledgement: This research was funded the EIT Raw Materials through the Innovative geophysical logging tools for mineral exploration - 16350 InnoLOG Upscaling Project.

How to cite: Bell, A., Del-Blanco, C. R., Jaureguizar, F., García, N., and Jurado, M. J.: Mineral interpretation results using deep learning with hyperspectral imagery, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-19667, https://doi.org/10.5194/egusphere-egu2020-19667, 2020

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 2 – uploaded on 06 May 2020
Version description: Hello again, The first and the last slide have been modified to have the authors with the first name and one surname,[...]
  • CC1: Comment on EGU2020-19667, Aris Lourens, 07 May 2020

    Dear authors,

    Thank you for presenting your nice work. I have two questions. 

    1) In slide 4 you suggest that you applied your method on hard meterial, which yields some problems in the recognition as shown in the pictures (1),(2),(3). Did you apply the method on soft sediments too? In that case you can have a flat cut of your core.

    2) You mention that you can apply your method in-situ. Does this mean that make observations/recordings in a borehole?

    Best regards,

    Aris Lourens

    • AC1: Reply to CC1, Andrés Bell, 07 May 2020

      Hello Aris,

      Thank you for your interest and your positive valuation. I am going to answer your questions:

      Concerning the first question, we have only applied our method on hard material because in the used database there are no soft sediments. However, our system could be able to work with them as well.

      With respect to the second question, yes, as it is our main objective, the proposed method can perform mineral recognition in-situ, in real-time, in the own borehole, to speed up the analysis of minerals and to make it more efficiently.

      For the acquisition process, there are several approaches: one of them is to use a borehole logging tool to take the hyperspectral data (this tool has been also developed in the project Innolog, in which we have been working, from EIT Raw Materials). Instead, by using a hyperspectral portable logging tool, cores can be extracted, and scanning can be performed in-situ. Therefore, the data acquisition would be performed in parallel with the real-time analysis.

      I hope that I have clearly answered your questions. As always, we are glad to hear further comments, questions or doubts.

      Best regards,

      The authors.

Display material version 1 – uploaded on 02 May 2020
  • AC1: Comment on EGU2020-19667, Andrés Bell, 02 May 2020

    Hello everyone, I am author of the presentation associated to EGU2020-19667 and we are open for any suggestions, doubts and feedback. This presentation is aimed to put in context and to explain the results we have obtained in automatic mineral recognition in hyperspectral imagery.

    Thank you for your attention and we wait patiently for any comments.

    Best regards,

    The authors.