Effect and correction of segregation error in coal size distribution estimation by image analysis on a conveyor belt

Abstract

Online monitoring of rock size is of great significance for mining and mineral processing, and machine vision is regarded as the most suitable approach. Segregation error of coal piles is inherent in image-based size distribution estimation, leading to the difference between surface size distribution and overall size distribution. In this research, we analysed the segregation error and proposed a correction method for image-based size distribution estimation of coal piles on a conveyor belt. The paint spraying of rock particles makes the compare of actual surface size distribution and overall size distribution possible. Rosin-Rammler particle characteristic equation was used to build the correction model to reduce the segregation error in size distribution estimation. Test results indicated that the corrected overall size distribution is close to the actual overall size distribution, and the maximal absolute error reduces from 19.98% to 2.04%. This investigation is useful and valuable for the optimization and improvement of image-based size distribution estimation of coal piles on a conveyor.