Automated Image Analysis and Contiguity Estimation for Liquid Phase Sintered Tungsten Heavy Alloys

TL;DR

This paper presents an automated image processing software for analyzing microstructure images of liquid phase sintered tungsten alloys, accurately estimating parameters like contiguity with minimal human intervention.

Contribution

The study introduces a novel automated method for microstructural analysis of tungsten alloys, reducing manual effort and providing open-source software adaptable to various microstructures.

Findings

Automated model estimates contiguity with 2.9-5.6% error.

The software accurately determines tungsten particle size and phase content.

The method is validated on two different alloy images.

Abstract

In this study an automated software model using digital image processing techniques is proposed for extracting the image characteristics and contiguity of liquid phase sintered tungsten heavy alloys. The developed model takes a typical image as input and processes it with no human intervention and provides the corresponding image characteristics and contiguity value. The image processing algorithm includes segmentation, binding point extraction, phase connection, particle count and contiguity estimation stages. For the output, microstructural parameters such as tungsten particle size, amount of tungsten phase and contiguity are determined. The model is implemented by using 6 different scanning electron microscope images of liquid phase sintered 90W-7Ni-3Fe and 93W-4.9Ni-2.1Fe allloys. The results indicate that relative to the manual measurements, the automated model can correctly estimate the contiguity with an error in the vicinity of 5.6% - 2.9% for these two alloys. The developed software can easily be adapted to be used for other microstructures. It is also provided as open-source and available for other researchers.