Flotation froth characterisation by closed domain (bubbles) colour analysis

Several procedures devoted to characterize the flotation froth structure and color have been developed, with the aim of contributing to better model a so complex process adopting a machine vision system based approach. In this paper a procedure based on the evaluation of the color domains characteristics belonging to the areas identified by the bubbles outlines resulting from segmentation techniques was designed and the results achieved were discussed. Possible solutions were proposed, developed and applied. The color analysis was performed starting from RGB (red, green, blue) surface froth images directly acquired in the flotation plant. The parameters resulting from the imaging techniques were correlated with those deriving from chemical analysis of the flotation products present in the froths expressed as Cu (%), Zn (%), Pb (%) and MgO (%) essay. The correlation was performed both considering the color of all the pixels belonging to the bubbles and just the color of the remaining pixels after the application of a "filtering operation". Good correlation was found in most cases considering the color of all the pixels; anyway better results were obtained in some cases introducing the "filtered" color parameters, showing a higher correlation coefficient in the regression analysis