Natural rock images classification using spatial frequency measurement

Texture has been widely accepted as a feature of primary significance in digital image processing field. In this paper, we present a simple approach to classify types of rock into three groups, which are homogenous, non-homogenous dark rock or bright rock containing less finer black grains, and non-homogenous bright rocks which contain numerous finer black grains or large black grains. Modified Spatial Frequency Measurement (SFM) is adopted to analyze and classify the textural rock feature. Prior to analysis process, the color images of rock are transformed to one dimensional pixel intensity array using Principal Component Analysis (PCA). This reduces the dimension of the data set to a spectral band, which capture almost all of the energy in the original textures. From the results of experiments, we demonstrate that using PCA and SFM to classify types of natural rock images yields promising result. Therefore, our proposed approach is a simply practical method to be used to classify different types of natural rock images led to the rapid decision-making.