Rock Classification Based on Images Color Spaces and Artificial Neural Network

For a fast and flexible access to the rock classification technology based on features extracted from rocks images, we propose a combination method to classify the rock type automatically with the images of core thin sections. The elements of feature space are from color and morphology features of rock images, and constructed with the statistical analysis result of standard arithmetic value into different color spaces. The relationship between feature space and rock type can be access with neural network. 1000 images from Ordos basin are used to test the availability and reliability of this method. Testing result shows this automatic rock type classification method get over 95.0% accuracy, which presents good prospect in practical usage.