Abstract—Estimation of the amount of fines in images of mineral particles using standard segmentation approaches is difficult. In this paper, an approach based on multivariate image analysis is presented for estimation of the amount of fines in particles on conveyor belts. The approach is based on two-level wavelet decomposition and morphological image operations, followed by feature extraction from gray level co-occurrence matrices. These features could be used with a simple k nearest neighbour model to estimate the proportion of fines in particulate images. Experimental results with coal and iron ore particles show that the performance of the method can yield better results than those achievable with standard methods.
Index Terms—Multivariate image analysis, particle size distribution, wavelets, textural feature extraction.
Anthony Amankwah is with the Department of Process Engineering, University of Stellenbosch, South Africa, Private Bag X1, Matieland 7602 (E-mail:firstname.lastname@example.org). Chris Aldrich is with the Department of Process Engineering, University of Stellenbosch, South Africa, Private Bag X1, Matieland 7602 (Tel +27 021 808 4485, Fax: +27 021 808 2059, E-mail: email@example.com).
Cite: Anthony Amankwah and Chris Aldrich, "Estimation of Particulate Fines on Conveyor Belts by Use of Wavelets and Morphological Image Processing," International Journal of Machine Learning and Computing vol. 1, no. 2, pp. 132-137, 2011.