OPTIMIZING DIGITAL IMAGE ANALYSIS OF THIN SECTIONS FOR RELIABLE PORE NETWORK CHARACTERIZATION

In this study, we present a digital image analysis methodology that applies machine learning, optimized for image processing and classification of thin section images for reliable pore network characterization. The methodology was applied to Upper Jurassic Jubayla Formation carbonate cores that are depositionally equivalent to the lower part of the super-giant Arab-D reservoirs found in Arabia. We find that the choice of image segmentation method has a significant impact on the final digital rock analysis results. The supervised machine learning method Support Vector Machine (SVM) performed the best in segmenting the macro-pores in the RGB thin section images compared to Random Forest and K-Means Cluster methods. 2D to 3D reconstruction by Multi Point Statistics (MPS) effectively reproduced the connectivity of the macropores in the studied rock sample. Pore size distribution and permeability calculated from the extracted pore network model matched well with the laboratory-measured data.