MicroCT and 3D Image Processing and Analysis to Investigate Strain-Hardening Cement-Based Composites (SHCC)

Abstract

AbstractX-ray micro-computed tomography (microCT) is a non-destructive technique that can provide 3D images of the internal microstructure of a composite material. Optimizing the analysis with modern computational tools leads to a higher precision in quantitative analysis and, consequently, to more accurate results. In this scenario, machine learning has been widely used as solutions for complex image processing and analysis tasks. The SHCC microCT images can be considered complex, given the small scale of analysis and the typical resolution of common microCT, as well as the small differences among the material constituents in terms of density and x-ray absorption. The present work brings innovative solutions for fiber and pore quantification in SHCC using Machine Learning. SHCC were tested in an in-situ testing device coupled to a microCT and the material mechanical response was correlated with microstructure changes through an image sequence. The internal displacement and strain were calculated by Digital Volume Correlation (DVC). The strain results were correlated with the initial quantification of the constituent phases of the material.KeywordsSHCCMicroCT3D image processing and analysisDeep learningDVC