About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Novel Strategies for Rapid Acquisition and Processing of Large Datasets From Advanced Characterization Techniques
|
| Presentation Title |
Conditional Diffusion Models for Microscopy Modality Transfer of Electron Backscattering Microscopy Diffraction Misorientations Maps from Optical Microscopy |
| Author(s) |
Nicholas Amano, Martin Müller, Bo Lei, Frank Mücklich, Elizabeth A Holm |
| On-Site Speaker (Planned) |
Nicholas Amano |
| Abstract Scope |
Texture analysis is a fundamental step in uncovering processing-structure-properties-performance relationships, motivating significant investment in the imaging of structured materials. In this study, we present a conditional diffusion model approach to synthesize electron backscattered diffraction (EBSD) inverse pole figure misorientation maps of quenched and tempered steel from light optical microscopy (LOM) micrographs. By leveraging the accessibility of LOM, our approach aims to accelerate the microstructural characterization process in steel production. This is enabled by a dataset of spatially aligned LOM and EBSD misorientation images acquired at the same sample regions and magnifications. We demonstrate that diffusion models can plausibly reconstruct visually and structurally consistent EBSD maps. While the generated maps cannot be used to extract the absolute misorientation values, we demonstrate statistically accurate texture prediction. Expanding this work to polarized LOM is an active area of research, offering the potential to further enhance the microstructural information contained within the optical data. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Characterization, Iron and Steel |