A technique for automatically segmenting images of the surface froth structures that are prevalent in industrial flotation cells

Abstract

The froth structures convey considerable visual information on cell performance, particularly in terms of grade and recovery. Fundamental to the design of a machine vision system for monitoring the operation is the ability to identify and extract those visual features that are descriptive of the surface froth. The individual bubbles give rise to complex 3-dimensional structures that are problematic to segment accurately and reliably. This paper briefly discusses the classical segmentation of surface froth images and then presents a morphological image processing technique that is capable of segmenting surface froth images accurately and reliably.<>