Recursive aggregate segmentation by erosion and reconstitution (RASER) to characterize concrete microstructure using complementarity of X-ray and neutron computed tomography

Abstract

The reconstruction of concrete microstructure with computed tomography (CT) images is a challenging problem because concrete is associated with poly-mineral aggregates, various aggregate sizes, high volume fraction of aggregates, rough aggregate surface, small and large connected aggregates, and others. To tackle some of the issues in concrete microstructure reconstruction, CT images are obtained using the complementarity of X-ray and neutron while a recursive aggregate segmentation by erosion and reconstitution (RASER) process is proposed to separate connected aggregates for both 3D and 2D. The RASER algorithm is based on the recursive and iterative processes of erosion and reconstitution of aggregate surface while increasing the eroding layer thickness until a constant number of aggregates is reached. Based on the RASER technique, the 3D segmented concrete microstructure illustrates that aggregates are not skewed based on their sizes and locations while neither under- nor over-segmented in this study. Furthermore, we found that most 2D microstructure analysis can lead to over-segmentation because of artificial section segmentation and rough aggregate surface.