资源环境科技发展态势分析平台

Modeling flow and transport in porous media using pore-scale modeling is reliant on rock properties derived from digital rock images using segmentation techniques. These digital rock images obtained using computed tomography incorporate the variation in the intensity of phases depending on the attenuation of X-rays. A standard technique is the segmentation of tomographic images based on user-selected grayscale thresholding, allowing the identification of different phases. This threshold is subjective based on the operator and results in loss of essential information about the grayscale variation after segmentation. This paper implements the gray-level co-occurrence matrix (GLCM) incorporating the full range of grayscale information. The GLCM captures the relative occurrence of grayscale values in a spatial map. These maps show visually connected/disconnected populations of different phases such as pore space, quartz grains, minerals, and other features. We show that each rock has its own GLCM signature depending on the variations in gray-level intensities. Several statistical measures are calculated: (1) GLCM contrast describing local variation in the gray-level intensities, (2) GLCM angular second moment, describing the rock homogeneity; (3) GLCM mean, describing weighted average of the probability of occurrence of features based on their location on the GLCM map; and (4) GLCM correlation, measuring the linear dependencies of grayscale values and the degree of (an) isotropy in the micro-computed tomographic images of each of the rock types. The GLCM method provides a pathway to alleviate user biases and allow automation of micro-computed tomography analyses.

Plain Language Summary The flow of fluids through a porous rock is heavily dependent on the geometrical structure of the rock. Modern techniques allow us to study the rock structure using high-resolution images obtained from computed tomography scanning. These high-resolution images represent different features of rocks by a range of grayscale values. The current interpretation method is to select a given grayscale value as a threshold to distinguish features from one another. This leads to user-biased outcomes, and also, important information about minerals is lost. Herein, we show that this problem can be alleviated by applying computer vision techniques directly to the original X-ray images and extracting directional, spatial, and frequency based information. This new perspective provides a fully automated analysis of rock characteristics and allows feature identification directly from the grayscale statistics. This contribution shows the descriptive power of the technique by characterizing rock structure using automated pattern recognition techniques.