Califice, Arnaud[Université de Liège - ULg > Département Argenco : Secteur GeMMe > Géoressources minérales & Imagerie géologique >]
Pirard, Eric[Université de Liège - ULg > Département Argenco : Secteur GeMMe > Géoressources minérales & Imagerie géologique >]
Proceedings IAMG 2010
29 august 2010 - 2 september 2010
Int Association Mathematical Geosciences
[en] microscopy ; grain size ; grain orientation
[en] Microtextural analysis of rocks has been addressed by several authors as an essential means to better understand the natural genesis of the material. But, it is also of paramount importance to those who try to predict the geotechnical or industrial behaviour of a rock under many forms of solicitation (mechanical, thermal, etc.). Quantitative modal (phase) analysis using point counting has already been discussed in depth by authors such as Chayes more than fifty years ago. Nowadays, automated image analysis with millions of pixels is easily available and improves statistical accuracy provided the classification step is correctly performed. Spitefully the assignment of a pixel to a given mineral phase or to a given crystal is often poorly satisfactory and remains the bottleneck of a fully automated textural analysis. Methods using a manual rotation of a polarizer in transmitted light microscopy have been developed by Starkey and Samantaray (1993) and further automated and improved by Fueten (1997). These allow to better delineate individual crystals in a thin section due to contrast in birefringence.
In this paper a similar technique using multiple orientations of a polarizer in reflected light microscopy has been used to contrast individual crystals in carbonated rocks. The maximum and minimum grey levels registered for each pixel allow for computing a bireflectance image whose variance is a good indicator of the misalignment of cristallographical orientations in the section. Moreover, the maximum of the reflectance gradient obtained for each orientation generates a good image of the grain boundaries and the presence of pores. This last one is quantitatively analysed using the intercept method to estimate the mean and variance of the grain size distribution. The paper presents a quantitative comparison of several different microtextures.