Scalable morphological accessibility of complex microstructures - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2021

Scalable morphological accessibility of complex microstructures

Johan Chaniot
Maxime Moreaud
Loïc Sorbier
Jean-Marie Becker
  • Fonction : Auteur
  • PersonId : 971954
  • IdRef : 070548285
Thierry Fournel

Résumé

This paper addresses the descriptors-based characterization of dense 3D microstructures using the unifying concept of accessibility, mixing local shape features with global topology. Underlying percolation and constrictivity features are jointly considered by probing the connected components of the microstructure with structuring elements with increasing sizes. Adapted morphological operations are combined to provide a scalable protocol embedding suitable descriptors applied on accessible volumes, yielding a sharp discrimination power. The suggested framework named A-protocol can efficiently analyze complex microstructures by applying a stratified sampling for the selection of paths’ endpoints, when connected. It stops when percolation ends, at a critical radius value. The A-protocol is tested on Cox multi-scale Boolean models using the Euler number as an arbitrarily chosen embedded descriptor. This computational protocol is available in the open access software environment plug im!.
Fichier principal
Vignette du fichier
Chaniot_2021_ScalableMorphologicalAccessibilityOfComplexMicrostructures.pdf (4.44 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03543639 , version 1 (22-06-2021)
hal-03543639 , version 2 (14-12-2021)
hal-03543639 , version 3 (15-12-2021)
hal-03543639 , version 4 (26-01-2022)

Identifiants

  • HAL Id : hal-03543639 , version 1

Citer

Johan Chaniot, Maxime Moreaud, Loïc Sorbier, Jean-Marie Becker, Thierry Fournel. Scalable morphological accessibility of complex microstructures: (Before review). 2021. ⟨hal-03543639v1⟩
119 Consultations
107 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More