Experimental Methods for Image Analysis of Two-Phase Metallic Microstructures

Abstract

In many engineering alloys, there are two phases at room temperature: alpha and beta. Traditionally, the process of categorizing visual features of the microstructure by phase has been performed manually according to ASTM International (or other) standard methods for counting and measuring microstructural elements. The benefit of this approach is that humans can recognize phase patterns relatively easily. Still, the work is tedious and puts practical limits on the quantity of data available for analysis. However, much current research involves digital image processing. Many of the current automated methods can easily create large amounts of data but sometimes at the cost of quality because almost all automatic processes to improve clarity, vary brightness, or highlight edges reduce the amount of information in the image. Here, the researchers employed a combination of machine learning and standard image processing techniques to provide large quantities of high-quality data. By training a segmentation classifier on features from several images, the researchers could delineate between the alpha and beta phases with higher accuracy than previously used operations and collect the size, shape, and orientation data desired.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 2021
Accession Number
AD1144104

Entities

People

  • Brady Butler
  • James Paramore
  • Jon-erik Mogonye
  • Nathaniel Bass
  • Trevor Hastings

Organizations

  • United States Army

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Abstracts
  • Department Of Defense
  • Digital Image Processing
  • Digital Images
  • Grain Size
  • Image Processing
  • Images
  • Information Operations
  • Machine Learning
  • Materials
  • Microstructure
  • Military Research
  • Standards

Readers

  • Computer Science/Computer Engineering/Data Science/Digital Signal Processing.
  • Neural Network Machine Learning.
  • Powder metallurgy of Titanium alloys.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks