Coal and rock classification with rib images and machine learning techniques

The classification of rock and coal will assist in automated coal rib rating and shearer horizon control, and is studied with machine learning techniques in this work. A database of rock and coal images is created by filtering photographs taken by researchers from the National Institute for Occupational Safety and Health (NIOSH). The classifier was trained with patches extracted from the coal and rock images, and an accuracy score of 0.9 was obtained. The trained classifier was then applied to classify rock from a new coal rib image with three rock layers of different thicknesses, and good agreement was achieved. The results demonstrate that it is promising to use machine learning techniques and rib images for rock and coal classification.