About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
2026 Technical Division Student Poster Contest
|
| Presentation Title |
SPU-34: A Machine Learning Based Approach for Quartz Texture Analysis Via Polarized Light Microscopy |
| Author(s) |
Alexander Alexey Kitchen, Brian Kendall, Tekle Khmiadashvili, Katalyn Denby, Jeffrey Rahl, Matteo Seita, Mengying Liu |
| On-Site Speaker (Planned) |
Alexander Alexey Kitchen |
| Abstract Scope |
Quartz is among the most common rocks: texture analysis provides geo-science data. From anisotropic optical property, grain orientations can reflect/transmit various light and intensities when illuminated with polarized white light. Our research uses polarized reflective microscope (PRM) to observe a rotating quartz sample, gathering intensity variations in red, green, and blue (RGB) channels pixel-by-pixel. We found the grain's c-axis orientation (inclination and azimuth angles from Electron-Backscatter-Diffraction) relates the change RGB intensity and maximum intensity phase shift (former non-linear, the latter discrete). Machine learning can correlate RGB minimum/maximum intensity to c-axis inclination through neural networks; we correlated the phase shift to its azimuth, with line of best-fit through each discrete band's gaussian center. Analysis indicates inclination and azimuth at 75% and 71.46% of predictions within +/-10 degrees, respectively. Machine learning PRM reduces analysis time from days to minutes. |
| Proceedings Inclusion? |
Undecided |