Semi-automated iron ore characterisation based on optical microscope analysis: Quartz/resin classification

Abstract

The present study focused on overcoming the primary problem faced by any quantitative mineralogical study involving iron ore characterisation using a reflected light optical microscope; distinguishing the quartz mineral from the epoxy resin in digital images taken from mounted polished sections. Difficulties arise in this case because both phases reflect in the same colour intensity range. To overcome this problem, a digital image analysis system denominated Opt-Lib was developed. In order to evaluate the system responsivity, a characterisation study and modal and liberation analyses were performed using typical Brazilian iron ore containing quartz and the main iron oxide/hydroxide minerals: magnetite, hematite, and goethite. For the system performance evaluation, these results were compared with those generated by the scanning electron microscopy (SEM)-based Mineral Liberation Analyzer for the same sample. The results show that the main advantage of the Opt-Lib system over the SEM-based system is that it facilitates differentiation, classification, and quantification of not only the quartz mineral, but also the iron oxide/hydroxide minerals within the sample, thus providing a more precise qualitative response.