Mineral Liberation Prediction - Contribution of Image Analysis Based on Mathematical Morphology

The increasing exhaustion of raw materials entails the exploitation of poorer orebodies where the target mineral occurs frequently in a very fine dissemination. Consequently it is necessary to grind more and more fine to achieve an adequate mineral liberation. In other words, it is of fundamental importance to know what size one should grind in order to avoid the following detrimental situations: 1. Excessive grinding - giving rise to the generation of ultra-fine slime particles which may be lost in the ulterior concentration process and simultaneously leading to higher energy consumption. 2. Incomplete grinding - giving rise to insufficient liberation and consequently leading to considerable losses of the valuable components in the tailings. Although this issue constitutes one of the most important steps in mineral processing, little research has been done compared to other processes, mainly due to the fastidious experimental work required. This paper describes a complete and integrated system of image analysis, where some algorithms to determine automatically the degree of li beration, for different levels of grinding, are proposed. These algorithms, mainly based on mathematical morphology theory were tested with images obtained from polished sections of different grain size fractions. Cases studies on a complex sulphide ore from an important copper and tin deposit located in the South of Portugal are discussed. Samples of the unbroken ore are also studied by image analysis to predict the degree of liberation obtained according to the fineness of grinding.