Semi-automated porosity identification from thin section images using image analysis and intelligent discriminant classifiers

  • Authors:

  • Javad Ghiasi-Freez;Iman Soleimanpour;Ali Kadkhodaie-Ilkhchi;Mansur Ziaii;Mahdi Sedighi;Amir Hatampour

  • Affiliations:

  • Faculty of Mining, Petroleum and Geophysics, Shahrood University of Technology, Shahrood, Iran;Ecole Nationale Superieur de Geologie, INPL University, Nancy, France;GeologyDepartment, Faculty of Natural Science, University of Tabriz, Tabriz, NW Iran, Iran;Faculty of Mining, Petroleum and Geophysics, Shahrood University of Technology, Shahrood, Iran;Faculty of computer engineering, Shahrood University of Technology, Shahrood, Iran;Faculty of Chemical Engineering, Dashtestan branch of Islamic Azad University, Dashtestan, Iran

  • Venue:
  • Computers & Geosciences
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Identification of different types of porosity within a reservoir rock is a functional parameter for reservoir characterization since various pore types play different roles in fluid transport and also, the pore spaces determine the fluid storage capacity of the reservoir. The present paper introduces a model for semi-automatic identification of porosity types within thin section images. To get this goal, a pattern recognition algorithm is followed. Firstly, six geometrical shape parameters of sixteen largest pores of each image are extracted using image analysis techniques. The extracted parameters and their corresponding pore types of 294 pores are used for training two intelligent discriminant classifiers, namely linear and quadratic discriminant analysis. The trained classifiers take the geometrical features of the pores to identify the type and percentage of five types of porosity, including interparticle, intraparticle, oomoldic, biomoldic, and vuggy in each image. The accuracy of classifiers is determined from two standpoints. Firstly, the predicted and measured percentages of each type of porosity are compared with each other. The results indicate reliable performance for predicting percentage of each type of porosity. In the second step, the precisions of classifiers for categorizing the pore spaces are analyzed. The classifiers also took a high acceptance score when used for individual recognition of pore spaces. The proposed methodology is a further promising application for petroleum geologists allowing statistical study of pore types in a rapid and accurate way.