Use of reflection flare spots for automatic recognition of froth parameters in potassium ores flotation

The paper defines the problem of automatic recognition of froth parameters in potassium ores flotation. A flotation engineer visually appraises froth in flotation cell and periodically changes flotation parameters. The problem consists in the factor of possible human error in flotation process control. The existing systems of froth product recognition in flotation cells, as well as bubbles images recognition methods are analyzed. Their main drawback — high requirements to image quality, is shown, demanding special conditions for camera filming, and accessories, unacceptable in potassium ores flotation. In order to solve the problem of froth color determination, two algorithms for image color profile definition have been developed. Performability of each algorithm is proved. A hypothesis is put forward about a relation between bubbles interflares distance and half sum of bubbles’ diameters. Image flares enhancing algorithm has been developed. Image binarization methods were analyzed, and a most effective one was chosen. Flare center determination algorithm is proposed. Software has been developed for statistical distribution of bubble sizes and mineral disseminations upon froth surface. It was checked through comparison of the software calculation results in four randomly chosen low-quality photographs with hand bubbles calculation results in the photograph. A close agreement of the results is demonstrated. In summary, over time duration within 0.8 s on a photograph of 72 dpi quality, the developed software creates statistical distribution of bubble sizes with a probability of 75 %, with an error within 10 %. This permits to use the described method for automatic recognition of froth parameters in flotation process automatic control or with a view to support decisions taken by a flotation engineer.

Hali tarjima qilinmagan