Modern electronic image-processing techniques have enabled mineral processing engineers to automate the determination of minerals in ore samples. The automatic recognition and quantification of minerals by light microscopy is one of the most important problems in ore-processing systems because determining the amount and degree of liberation of the constituent minerals in ore is necessary for further processing. Measurement of the size and liberation degree of minerals is also required for automatic control of the grinding process. This paper suggests an automated method for segmenting and quantifying the size and amount of minerals in ore using micrographic images. A simple normalized colour-based statistical segmentation method is proposed to exploit the average value, standard deviation and distribution of RGB colour components of mineral patterns in an ore image. The method also determines the deviations of colour components of the minerals to improve the segmentation. A Naive Bayes classifier is also introduced for segmenting the minerals. The performance of method is examined using micrographs in variety of qualities. The method performs segmentation accuracy over 90%. Additionally, the success rates of the methods were found to be over 85% in measuring the grain sizes in a ground sample, and 86% in measuring liberation degrees of minerals after grinding process. (c) 2012 Elsevier Ltd. All rights reserved.