Premium
Application of highlight removal and multivariate image analysis to color measurement of flotation bubble images
Author(s) -
Yang Chunhua,
Xu Canhui,
Gui Weihua,
Zhou Kaijun
Publication year - 2009
Publication title -
international journal of imaging systems and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 47
eISSN - 1098-1098
pISSN - 0899-9457
DOI - 10.1002/ima.20208
Subject(s) - inpainting , artificial intelligence , computer science , froth flotation , computer vision , specular reflection , bubble , multivariate statistics , image (mathematics) , pattern recognition (psychology) , materials science , optics , machine learning , physics , parallel computing , metallurgy
Machine vision based analysis provides a novel technology for froth flotation monitoring. Froth images collected are characterized by fully occupied bubbles with different size and shape under various illuminations. Convex bubbles lead to the formation of white spots that seriously affect froth color measurement. In this article, specular highlights are detected and preprocessed so as to estimate underlying color of white spots region. Because of the fact that color information is believed to be related to flotation performance, therefore, after the application of highlight inpainting, multivariate image analysis is proposed to extract color features, which are further related to mineral grades by a orthogonal least square regression model. The established relationship provides a promising empirical model to predict mineral grade, which is a significant indicator for flotation performance. Experimental results show that, when compared with traditional methods, the proposed algorithm can achieve a robust color measurement and predict mineral concentration effectively. © 2009 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 19, 316–322, 2009