Gaussian segmentation of BSE images to assess the porosity of concrete

  • Authors:

  • D. J. Robinson;F. Murtagh;P. A. M. Basheer

  • Affiliations:

  • School of Civil Engineering, Queen's University of Belfast, Northern Ireland;School of Computer Science, Queen's University of Belfast, Northern Ireland;School of Civil Engineering, Queen's University of Belfast, Northern Ireland

  • Venue:
  • ICCST '02 Proceedings of the sixth conference on Computational structures technology
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a model-based Gaussian clustering approach for the identification of the porosity, hydrated and unhydrated phases constituting a backscatter electron image of the microstructure of cement paste. The likelihood function for the Gaussian mixture is optimised using the expectation-maximization algorithm and the quality of fit is assessed using the Bayes information criterion. The technique provides a consistent and repeatable means of phase identification within the microstructure of cement paste. The application of the approach has been demonstrated through an example.