LEGOMAT: Locally Extracted Globally Organized Microstructure Models using Markov Random Fields

Abstract

University of Michigan propose theory and software for building three dimensional (3D) microstructural maps of engineering components through Markovian inference from two dimensional (2D) measurements. Here, most of the microstructures will be generated synthetically based on a small set of microscopic scans. The synthetic microstructures will be generated using an inference algorithm based on Markov random fields (MRFs) [11] and the microstructures will be used to fill in gaps in information at the component scale.

Document Details

Document Type
DoD Grant Award
Publication Date
Apr 09, 2018
Source ID
FA95501810091

Entities

People

  • Veera Sundararaghavan

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Michigan

Tags

Readers

  • Computer Vision.
  • Mathematical Modeling and Probability Theory.
  • Powder metallurgy of Titanium alloys.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms