Rock lithological classification using multi-scale Gabor features from sub-images, and voting with rock contour information

This article introduces an innovative method for remote lithological classification in mining plants, a critical aspect for determining rock size and grindability, directly impacting the control of the grinding process. The approach utilizes a single digital video camera for capturing images of rocks on a conveyor belt, breaking each image into sub-images to extract texture information across various spatial scales. Gabor filters and a support-vector machine are employed for feature extraction and classification. The method’s efficacy was tested on three databases, revealing significant improvements in classification accuracy, ranging from 8.3% to 26%, compared to previous results. These findings underscore the potential of this method in enhancing mining plant operations.