About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Recent Advances in Electron Back-Scattered Diffraction and Related Techniques
|
| Presentation Title |
Sub-Grain Detection in EBSD Datasets Using Clustering Algorithms in Machine Learning |
| Author(s) |
Mrinmoy Chanda, Nilesh Prakash Gurao |
| On-Site Speaker (Planned) |
Mrinmoy Chanda |
| Abstract Scope |
The evolution of sub-grains (intragranular regions with misorientations less than 10-15°) plays an important role in deciding the plastic deformation, recovery, and recrystallization in polycrystalline materials. A thorough characterization of sub-grains from EBSD datasets remains a challenge as conventional misorientation-based segmentation methods often incorrectly classify noise as structure or overlook subtle orientation gradients. We determine pixel-wise orientation extraction, misorientation calculations with nearest neighbours, and visualisation of dislocation-induced substructures to implement and compare the Fast Multiscale Clustering (FMC) and Anti-Leak Grain Identification (ALGrID) approach in the Orix Python library. The former identifies orientation clusters with varying sensitivity level while the later reconstructs closed boundaries using a graph-based approach centred on disorientation paths and is more effective at minimising false segmentation and capturing connected structures. This integrated framework improves the accuracy and automation of sub-grain segmentation, facilitating quantitative analysis of GND-driven microstructural evolution and texture development. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, |