Automation of intercept method for grain size measurement: A topological skeleton approach

Abstract

In the microstructure characterization of metallic materials, the intercept method is one of the most widely accepted approaches to determine average grain size due to its simplicity, accuracy, and the ability to handle both equiaxed and non-equiaxed grain structures. However, its manual implementation is relatively time-consuming and error-prone, and the design of automated implementations is challenging due to the requirement of recognizing, classifying, and scoring different types of intersections (between test patterns and grain boundaries) by international standards such as ASTM E112 and EN ISO 643. In this research, a novel algorithm is proposed to automate the intercept method for grain size measurement from microscopic images. Building on topological skeletons, the algorithm is able to extract continuous and closed grain boundaries from the raw image, and determine the average grain size by recognizing and classifying different types of intersections in accordance with international standards. The effectiveness and efficiency of the proposed algorithm is validated on metallographic microscope images from both high-purity iron and stainless steel. Additionally, our algorithm has been extended to automate other standard grain size measurement methods such as the planimetric method and the whole grain area method.