GrainStat: A Python Framework for Automated Grain Microstructure Analysis in Materials Science

Abstract

Characterization of grain microstructure in materials science is a key area of investigation as grain microstructure can affect the mechanical properties, process variables, and performance in various engineering applications. Traditionally, characterizing grain microstructure is a time-consuming, subjective endeavor with many sources of variability in the analyses, especially when examining high-resolution micrographs and complex microstructure with multiple phases. This work introduces GrainStat, an automated Python framework for characterizing grain microstructure with advanced image processing, statistical analysis, and visualization functionality. The framework includes efficient algorithms for grain segmentation, morphological analysis, and statistical characterization while compliant with existing international standards (ASTM E112, ISO 643). Test applications indicate substantial efficiencies of analysis, reproducibility of measurements, and accuracy of representation when using the frameworks automatic processes when compared to traditional, manual methods show improvements of 10-100× processing time, and coefficient of variation of less than 2% for repeated measurements. The open-source framework acts as a resource for research and industry in materials characterization so that advanced microstructure analysis can be made accessible to all.