Coal and Rock Classification with Rib Images and Machine Learning Techniques (Mining, Metallurgy & Exploration)

Classification of rock and coal is one preliminary problem for fully automated or intelligent mining. It assists in automated rib stability analysis and enables the shearer to adjust the drums without human intervention. This paper studies the classification of rock from coal on rib images using machine learning techniques. A database of rock and coal images was created by filtering photographs taken by NIOSH researchers during site visits. Machine learning was conducted on patches, smaller images randomly extracted from each rock or coal image. After training, the classifier was validated with the testing dataset and achieved an accuracy score of 0.9.