Underground pre-concentration is an efficient stage in reducing the cost of mining operations. During this operation, extracted rocks are monitored underground and most of the waste rock is rejected while the ore is kept. In this project, image processing and pattern recognition algorithms will be employed to characterize different types of ores as well as rocks. This system will be part of a machine vision system that will be used for sorting ores from waste rocks in an underground preconcentration in a mine. The system uses a camera to take pictures of rocks as they slide on a conveyor belt. The system then uses features describing the color of rocks to categorize them into the correct category (e.g., ore / waste rock). The successful performance of this pattern recognition system will result in a significant amount of cost saving as well as reducing negative impacts on the environment.