Investigation of image segmentation effect on the accuracy of reconstructed digital core models of coquina carbonate*
Wang Hai-Tao 1,2,3, Wang Li 4,5, Lai Fu-Qiang♦1,2,3, and Zhang Jin-Yan 6
1. School of Petroleum Engineering, Chongqing University of Science & Technology, Chongqing 401331, China.
2. National and Local Joint Engineering Research Center of Shale Gas Exploration and Development, Chongqing 401120, China.
3. Key Laboratory of Complex Oil and Gas Field Exploration and Development, Chongqing University of Science &Technology, Chongqing 401331, China.
4. The First Oil Production Plant of Changqing Oilfi eld, PetroChina Company Limited, Shanxi Yan’An 716000, China.
5. Logging Application Research Institute, CNPC Logging, Shanxi Xi’ An 710000, China.
6. Well Logging Company, Shengli Oilfi eld Service Corporation, Sinopec, Shandong Dongying 257096, China.
Abstract：
Digital core models reconstructed using X-ray tomography (X-CT) enable the quantitative characterization of the pore structure in three dimensions (3D) and the numerical simulation of petrophysics. When the X-CT images accurately reflect the micro structures of core samples, the greyscale threshold in the image segmentation determines the accuracy of digital cores and the simulated petrophysical properties. Therefore, it is vital to investigate the comparison parameter for determining the key greyscale threshold and the criterion to describe the accuracy of the segmentation. Representative coquina digital core models from X-CT are used in this work to study the impact of grayscale threshold on the porosity, pore percolation, connectivity and electrical resistivity of the pore scale model and these simulations are calculated by Minkowski functions, component labeling and fi nite element method, respectively, to quantify the pore structure and simulate electrical resistivity. Results showed that the simulated physical properties of the digital cores, varied with the gradual increase of the greyscale threshold. Among the four parameters related to the threshold, the porosity was most sensitive and chose as the comparison parameter to judge the accuracy of the greyscale threshold. The variations of the threshold change the micro pore structures, and then the electrical resistivity. When the porosity of the digital core model is close to the experimental porosity, the simulated porosity exponent matches the experimental porosity exponents well. The good agreement proved that the porosity is the critical comparison parameter to describe the accuracy of image segmentation. The criterion is that the porosity of the digital core after segmentation should be close to the experimental porosity.
. Investigation of image segmentation effect on the accuracy of reconstructed digital core models of coquina carbonate*[J]. APPLIED GEOPHYSICS, 2020, 17(4): 501-512.
2020, Vol. 17 
