Hyper- and multispectral image analysis methods for mineral identification: two cases of study
This study briefly reviews geological applications of optical remote sensing and puts in practice some data analysis methods through two cases of study: ore assessment through the analysis of a hyperspectral image of a test panel of mine samples, and the recognition of mine waste materials through t...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | Consejo Superior de Investigaciones Científicas (CSIC) |
| Repositorio: | DIGITAL.CSIC. Repositorio Institucional del CSIC |
| OAI Identifier: | oai:digital.csic.es:10261/233456 |
| Acceso en línea: | http://hdl.handle.net/10261/233456 |
| Access Level: | acceso abierto |
| Palabra clave: | ASTER Mining Hyperspectral Remote Sensing Imaging spectrometry Mineral exploration Multispectral Remote Sensing Machine Learning. |
| Sumario: | This study briefly reviews geological applications of optical remote sensing and puts in practice some data analysis methods through two cases of study: ore assessment through the analysis of a hyperspectral image of a test panel of mine samples, and the recognition of mine waste materials through the analysis of a multi-spectral satellite image of a mining site (Sotiel, SW Spain). Principal Components and Linear Discriminant Analysis have been applied, as well as classification methods (Random Forest, Support Vector and Spectral Angle Mapper). We also have analysed and interpreted the spectral signatures. In the case of the panel of mine samples, Random Forest classification has been able to discriminate ore (cassiterite and wolframite) with a 96 % of accuracy, which puts in evidence the interest of developing “in situ” systems for the fast assessment of ore grade based on hyperspectral sensors. In the case of the mining area, our results have shown a high consistency of ASTER spectral signatures with the spectra of the endmember minerals previously derived by Spectral Mixture Analysis from aerial hyperspectral imagery by other authors, and a similar geographic distribution of major contaminated areas has been found. Water polluted by mine waste at different intensities in the mine ponds has been detected in the ASTER imagery. Combining the high analytical capabilities of aerial hyperspectral imagery with the extensive monitoring capabilities of satellite imagery can be effective for the control of mining environments. |
|---|