Traditional identification of ore minerals with reflected light microscopy relies heavily on the experience
of the observer. Qualified observers have become a rarity, as ore microscopy is often neglected in today’s
university training, but since it furnishes necessary and inexpensive information, innovative alternatives are
needed, especially for quantification. Many of the diagnostic optical properties of ores defy quantification,
but recent developments in electronics and optics allow new insights into the reflectance and colour
properties of ores. Preliminary results for the development of an expert system aimed at the automatic
identification of ores based on their reflectance properties are presented. The discriminatory capacity of
the system is enhanced by near IR reflectance measures, while UV filters tested to date are unreliable.
Interaction with image analysis software through a wholly automated microscope, to furnish quantitative
and morphological information for geometallurgy, relies on automated identification of the ores based on
the measured spectra. This methodology increases enormously the performance of the microscopist;
nevertheless supervision by an expert is always needed
Thesis Collection
Last Deposit
