Texture analysis of sandstone surfaces

This report is the contribution from the Norwegian Computing Center (NR) to the workshop Flat Image Texture Analysis, August 29, 1994, organized by SINTEF. The objective of the workshop was to study possible textural features that can be applied to discriminate between different sandstone facies. Due to limited resources, only relevant textural operator currently available at NR were tested. These are GLCM features, Simultaneous Autoregressive (SAR) models, Gabor filters, at set of six miscellaneous local texture operators. The available data set consisted of four samples only, so we have been cautious in drawing any final conclusions. The experiments in the different spectral bands and (2) some of the textural features. Of the four sandstone facies parameters grain size, porosity, and contents of quartz and feldspar, we found indications of relationships between gray levels and the textural features. A relatively high correlation was found between gray gray level and the textural features GLCM Correlation, SAR textural features, Contrast and Homogeneity. Other textural features seem to have some discriminative properties, but at least they cannot be used separately. Before work on further adaption or development of textural operators is initiated, we strongly recommend that these textural features are tested statistically in a data set consisting on a few hundred samples.