Prediction of microstructural representativity from a single image

  • 2025-07-28 12:43:01
  • Amir Dahari, Ronan Docherty, Steve Kench, Samuel J. Cooper
  • 0

Abstract

In this study, we present a method for predicting the representativity of thephase fraction observed in a single image (2D or 3D) of a material. Traditionalapproaches often require large datasets and extensive statistical analysis toestimate the Integral Range, a key factor in determining the variance ofmicrostructural properties. Our method leverages the Two-Point Correlationfunction to directly estimate the variance from a single image, therebyenabling phase fraction prediction with associated confidence levels. Wevalidate our approach using open-source datasets, demonstrating its efficacyacross diverse microstructures. This technique significantly reduces the datarequirements for representativity analysis, providing a practical tool formaterial scientists and engineers working with limited microstructural data. Tomake the method easily accessible, we have created a web-application,www.imagerep.io, for quick, simple and informative use of the method.

 

Quick Read (beta)

loading the full paper ...