Computer vision applications in mineral processing research

Abstract

The Tuscaloosa Research Center ofthe U.S.Bureau of Mines is utilizing computer-vision as a measurement tool and sensor in several of its research projects. These projects are directed at development of mineral processing technology with the goal of precise control of systems of increasing complexity. Adequate control of complex systems requires high precision sensory input. Computer vision hasproved to be aflexible tool for accurate and reproducible data-collection using digitized images. 'This paper describes the successful application of computer vision technology in four research projects at the Tuscaloosa Research Center. These projects include mineral identification in phosphate tailings, estimating chemical composition using color, and particle size and settling rate analysis during floc formation. A brief introduction to computer vision and a description of the hardware and software used is given and the importance of computer vision as a sensor is emphasized.for estimating chemical composition of solutions. Two other projects are described which fall under the general area of the study of floc formation, particularly settling rate and particle size analysis. Settling rates of fine particles are monitored by use of computer vision to track and record the fluid/particle interface. The flocculation project is directed at obtaining statistical information on floc characteristics such as size, density, settling, and growlh rates from images captured in near-real time during the formation process.The following discussion provides an overview of techniques and processes associated with computer-vision technology as applied to these applications. A. Computer Vision I. INTRODUCTIONComputer control, and thus automation, of minerals beneficiation is limited by a lack of adequate sensors 10 provide information on the current state of the process. Recent cost reduction of image acquisition hardware, software, and microcomputers permit the economical use of computer vision as sensory input in process control. Computer vision can often be substituted for a human operator's visual observations or for more complex and costly sensors. This sensory attribute has been put to good use in several projects at the Tuscaloosa Research Center which will be described in this paper.Computer vision, or machine vision as it is sometimes called, is thc construction of explicit, meaningful descriptions of physical objects from images. This definition implies image understanding, or description building, which is quite different from image processing [I]. However, image processing techniques must normally be extensively used in order to implement image understanding.Following a background discussion on computer vision, the applications are discussed beginning wilh a project h a t is directed at quantitative determination of the amount of the mineral apatite in phosphate tailings following flotation. A second project uses the camera as a color sensor, or colorimeter, 2013 0-7803-0634-1/92$03.00 OEEE I 1 -Computer vision had its origins ov...

Digital image processing (DIP) application on the evaluation of iron-rich heavy mineral concentrates produced from river sand using a sequential mineral processing approach

“…Therefore, such methods are used in many disciplines including medicine, geodesy, photogrammetry, earth sciences, chemistry, astrophysics, biotechnology, and metallurgy (Maras et al, 2010). Moreover, the DIP method has been used for the evaluation of several processes such as flotation (Moolman et al, 1994;Aldrich et al, 2010), particle size distribution analysis (Kursun, 2009), and solid-liquid separation (Stanford et al, 1992) in the mineral processing discipline. Alternatively, these methods have also been researched in several studies as an alternative to traditional techniques used in ore grade analyzes (Singh and Rao, 2006;Khorram et al, 2011;Perez et al, 2011).…”