Unsupervised clustering for Electrofused Magnesium Oxide sorting

This research is concentrated on using unsupervised learning technique and digital image processing to cluster mineral materials, Electrofused Magnesia Oxide specifically, for industry automation. We proposed a technique to construct an image database by generating data from images using a digital image process. This is based on a simple histogram mode and intensity deviation. A group of two popular clustering algorithms has been tested to develop an automatic system for industry. We have concluded that the best suited algorithm for this application in the mineral industry from this group of two algorithms is the k-means algorithm.