Application of pattern recognition methods to automatic identification of microscopic images of rocks registered under different polarization and lighting conditions

Abstract

The paper presents the results of the automatic classification of rock images, taken under an optical microscope under different lighting conditions and with different polarization angles. The classification was conducted with the use of four pattern recognition methods: nearest neighbor, k-nearest neighbors, nearest mode, and optimal spherical neighborhoods on thin sections of five selected rocks. During research the CIELAB color space and the 9D feature space were used. The results indicate that changing both lighting conditions and polarization angles results in worsening the classification outcome, although not substantially. During the automatic classification of rocks photographed under different lighting and polarization conditions, the highest number of correctly classified rocks (97%) is given by the nearest neighbor method. The results show that the automatic classification of rocks is possible within a predefined group of rocks. The results also indicate the optimal spherical neighborhoods method to be the safest method out of those tested, which means that it returns the lowest number of incorrect classifications.