Electrofused magnesium oxide classification using digital image processing and machine learning techniques

This research is focused on using digital image processing and machine learning techniques to classify electrofused magnesia for industry automation. We generate the data from different images by using a modern digital image process. This research proposes a new method to construct the digital image database. The proposed new method is based on simple histogram mode and intensity deviation. A group of six popular machine learning algorithms has been tested to build up an automatic system for industry. We have concluded that the best suited algorithm for magnesia industry automation from this group is the PART algorithm.