Chapter 8: Image analysis for advanced characterization of industrial minerals and geomaterials
E. Pirard and P. Sardini
This chapter is dedicated to the science of extracting quantitative information from digital images representing minerals and rocks. Because of the extraordinary complexity of natural textures, but also due to the wide diversity of mineral species, such analysis is still regularly performed by geologists using manual point-counting methods and basic stereological principles. If one aims to automate the process, it is essential to realise that images have to be acquired wherein individual minerals are contrasted as much as possible. This depends heavily on the quality of the imaging instrument and the attention dedicated to sample preparation.
After reviewing a range of modern mineral-imaging modes using electrons, X-rays, photons and even nanosized probes, the chapter focuses on the
tools and techniques most commonly used to archive and process digital images. Special emphasis is given to image segmentation techniques that allow the user to classify pixels and map homogeneous domains that might correspond to specific
minerals or single crystals.
The image analysis part sensu stricto addresses the quantitative description of mineral abundance before suggesting different techniques to analyse size and shape distributions of particles and grains. A brief introduction to network description is also given with special attention paid to the powerful concept of intercepts.
Finally, images are presented as support for physical simulations that bring new insight to behaviour of geomaterials with respect to processes such as diffusion and percolation.
Go to the table of contents for this book.
Go to the Mineralogical Society’s online shop to buy a copy of the book from which this chapter is taken