| Abstract Scope |
The microstructure of materials is a crucial factor in determining their macroscopic properties, including yield stress, hardness, fatigue, and creep resistance. Among various microstructural properties, grain size distribution plays a significant role in multiple physical relationships, such as the Hall-Petch effect. We propose a novel computer vision method that utilizes super-pixel segmentation to rapidly extract grain geometry information from grayscale micrographs, such as scanning electron microscopy (SEM) images. The proposed pipeline employs Quickshift, a super-pixel segmentation technique that groups perceptually similar pixels, followed by Region Adjacency Graph (RAG) Merging to address over-segmentation issues. This study demonstrates the validity and use cases of the proposed computer vision method in analyzing grain structure rapidly and efficiently, potentially saving researchers significant time and resources. |