3D multi-scale segmentation of granular materials

Abstract

On a microscopic scale, a pyrotechnic material is made of a polymer matrix containing grains with different sizes and shapes. Its physical behaviour can be predicted by homogenization. Information about the morphology of the grains can be obtained by different ways. One of these ways is 3D image processing. This has been made easier by the use of a new-imaging technique, the microtomography, allowing fast three-dimensional reconstruction and processing. In this paper images of two granular materials are segmented by means of 3D mathematical morphology algorithms, based an a multi-scale extraction of markers for watershed segmentation.