Spiral Concentrator Interface Monitoring through Image Processing: Optimization for Parameter Selection

Spiral concentrators are robust gravity separation devices that allow the concentration of slurry streams in terms of a desired mineral of interest. The optimal splitter position in a spiral concentrator is dependent on the interface position(s) between concentrate, middlings and/or gangue streams in the spiral trough. Image processing methods can be used to detect interface positions, which could be used in spiral concentrator operation monitoring and control. However, the algorithms required for interface monitoring by means of image processing involve a large number of parameters that must be specified. Such parameter specification can become onerous if done on a manual or ad hoc basis. The goal of this study is to investigate the potential of applying an optimization approach to determine optimal parameters for interface detection through image processing. Results are promising, indicating good interface detection, even for small training data set sizes (e.g. 50 images).