Rock Lithological Classification Based On Gabor Filters And Support Vector Machine

Estimation of rock composition in mineral plants is important to determine rock size and grindability which in turn may improve control of the grinding process. Variations in ore grindability and size distribution directly affects mills power consumption and throughput. Therefore, it is desirable to develop new tools to estimate rock lithological composition. In this paper, a new method for lithological classification based on machine vision is proposed. The method is based on a single video camera and employs information about texture using several different spatial scales. Texture information is extracted from rock images using Gabor filters that can be adjusted at different spatial scales and orientations. After feature extraction, rock images are classified using support-vector machine (SVM). The method was tested on a database containing sub-images of 64x43 pixels of five ore types as follows: massive sulphide (MS), disseminated sulphide (DS), net textured (NT), gabbro (G), and peridotite (P). These ore types were assigned to three grindability classes: soft (MS), medium (DS and NT) and hard (G and P). The database was divided in 2 subsets for training, cross validation, and for test. Classification accuracy is compared with previously published results on a dataset with three different classes of rocks according to their hardness. Results show that our proposed method reached significant improvements (between 13%-20%) in classification accuracy relative to previously published results. Keywords: rock classification, lithological classification, grindability estimation, Gabor feature extraction, texture features