Image-Based Rock Typing Using Specific Surface and Iterative Convolution-Thresholding Method

Abstract

Image-based rock typing (IBRT) is an effective way to understand the pore scale heterogeneity of the reservoir samples. IBRT is aimed at segmenting a rock sample’s image into different regions where each region represents a homogeneous porous medium, also known as rock type. Currently, the phase-field rock typing method has attracted more attention due to its impressive performance in classifying the heterogeneous rock images with highly irregular pore structures. In this paper, a modified specific surface CV (SSCV) model is proposed to realize the IBRT. In the SSCV model, the specific surface of a pixel is calculated within a given size neighborhood to distinguish different rock types, and the iterative convolution-thresholding method (ICTM) is applied as the classifier. Compared to the LHFCV method, an existing phase-field rock typing method, the proposed SSCV is capable of processing the images with more than two rock types and can be solved by ICTM which has higher computational efficiency. The proposed SSCV method has demonstrated remarkable performance in the segmentation of various images of both synthetic and natural rock samples.