Flotation Froth Image Analysis by Use of a Dynamic Feature Extraction Algorithm

• Published in: IFAC-PapersOnLine Vol. 49; no. 20; pp. 84 - 89
• Main Authors: Fu, Yihao, Aldrich, Chris
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 2016
• Subjects: Computer Vision; Flotation; Image Analysis; Local Binary Patterns; Machine Learning; Local Binary Patterns; Flotation; Image Analysis; Computer Vision; Machine Learning
• ISSN: 2405-8963; 2405-8963
• DOI: 10.1016/j.ifacol.2016.10.101

Froth image analysis has been well established as a means to infer the performance of froth flotation cells in real time. Apart from linking the appearance of the froth to the behavior of the flotation system, the dynamic behaviour of the froth is also an important determinant of the performance of the flotation cell, and ideally, this information should also be taken into consideration. In this investigation, the dynamic behaviour of the froth was incorporated implicitly in the features extracted from the images. As a case study, mineral mixtures consisting of realgar, orpiment and quartz were floated in a laboratory batch flotation cell. Videographic mages of the froths generated by the experiments and a dynamic local binary pattern algorithm (LBP-TOP) was used to extract features from the video data. A random forest model could subsequently be built to reliably classify the conditions prevailing in each of the batch runs. The dynamic LBP algorithm did not perform significantly better than its 2D equivalent that did not incorporate the temporal behaviour of the froth, as both approaches could very reliably identify the different froth classes.