Text only print
Full screen print
Seminar on Mathematics for various disciplines
Seminar information archive ～10/22｜Next seminar｜Future seminars 10/23～
| Date, time & place | Tuesday 10:30 - 11:30 056Room #056 (Graduate School of Math. Sci. Bldg.) |
|---|---|
| Organizer(s) | Yoshikazu Giga, Naoyuki Ishimura(Chuo University), Norikazu Saito, Masahiro Yamamoto, Hiroyoshi Mitake(Waseda University) |
| URL | https://www.math.sci.hokudai.ac.jp/coe/sympo/various/index_en.html |
2026/12/08
10:30-11:30 Room #056 (Graduate School of Math. Sci. Bldg.)
Prashant Athavale (Clarkson University)
Total Variation Methods for Denoising and Inpainting Crystallographic Orientation Maps (English)
Prashant Athavale (Clarkson University)
Total Variation Methods for Denoising and Inpainting Crystallographic Orientation Maps (English)
[ Abstract ]
Electron backscatter diffraction (EBSD) provides spatial maps of crystal orientations in polycrystalline materials. These measurements often contain noise and missing data, complicating the identification of grains and their boundaries. In this talk, I will introduce the mathematical challenges of restoring orientation maps and present our work on total variation methods for denoising and hybrid methods for filling missing regions. Particular attention will be given to preserving grain boundaries while recovering useful microstructural information. I will also discuss ongoing work on using shape tensors and statistical comparisons of grain geometry to guide stopping criteria for denoising flows. The presentation will emphasize connections between variational methods, geometric representations, and materials imaging.
Electron backscatter diffraction (EBSD) provides spatial maps of crystal orientations in polycrystalline materials. These measurements often contain noise and missing data, complicating the identification of grains and their boundaries. In this talk, I will introduce the mathematical challenges of restoring orientation maps and present our work on total variation methods for denoising and hybrid methods for filling missing regions. Particular attention will be given to preserving grain boundaries while recovering useful microstructural information. I will also discuss ongoing work on using shape tensors and statistical comparisons of grain geometry to guide stopping criteria for denoising flows. The presentation will emphasize connections between variational methods, geometric representations, and materials imaging.
