Geometry segmentation of voxelized representations of heterogeneous microstructures using betweenness centrality

Rui Yuan, Sudhanshu S. Singh, Nikhilesh Chawla, Jay Oswald

Research output: Contribution to journalArticlepeer-review

Abstract

We present a robust method for automating removal of “segregation artifacts” in segmented tomographic images of three-dimensional heterogeneous microstructures. The objective of this method is to accurately identify and separate discrete features in composite materials where limitations in imaging resolution lead to spurious connections near close contacts. The method utilizes betweenness centrality, a measure of the importance of a node in the connectivity of a graph network, to identify voxels that create artificial bridges between otherwise distinct geometric features. To facilitate automation of the algorithm, we develop a relative centrality metric to allow for the selection of a threshold criterion that is not sensitive to inclusion size or shape. As a demonstration of the effectiveness of the algorithm, we report on the segmentation of a 3D reconstruction of a SiC particle reinforced aluminum alloy, imaged by X-ray synchrotron tomography.

Original languageEnglish (US)
Pages (from-to)553-559
Number of pages7
JournalMaterials Characterization
Volume118
DOIs
StatePublished - Aug 1 2016

Keywords

  • Betweenness centrality
  • Clustering
  • Geometry segmentation

ASJC Scopus subject areas

  • Mechanical Engineering
  • Mechanics of Materials
  • Materials Science(all)
  • Condensed Matter Physics

Fingerprint Dive into the research topics of 'Geometry segmentation of voxelized representations of heterogeneous microstructures using betweenness centrality'. Together they form a unique fingerprint.

Cite this