Quantification of porous media using computerized tomography and a statistical segregation threshold

A statistical segregation threshold (SST) that determines the spatial distr ibution of solids and pores in computerized tomography (CT) images of porou s media is presented. The proposed algorithm combines basic image processin g techniques with consideration of measurement errors and CT scanning resol ution. While traditional threshold methods use only one or two parameters t o identify pore features in CT images, the SST characterizes the spatial di stribution of features by segregating image voxels into groups and treating each group differently according to its known statistical properties. This provides a major improvement when pores are smaller than the image resolut ion. By classifying the CT image into mixed-component, peak and tail-voxel fractions, the proposed algorithm provides a semiautomatic and theoreticall y sound technique for component segregation. The SST algorithm results are insensitive to the local threshold value and sample heterogeneity. Three ac rylic test objects with various void sizes and water content are used to te st and demonstrate the SST.