About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Characterization of Minerals, Metals and Materials
|
Presentation Title |
Non-local Means Denoising of EDS Spectra for Rapid Mapping of Composition in Nickel Aluminum Bronze |
Author(s) |
Patrick Gregory Callahan, Dillon S Watring, David J Rowenhorst |
On-Site Speaker (Planned) |
Patrick Gregory Callahan |
Abstract Scope |
The goal of this work is to enhance composition determination in materials by energy dispersive X-ray spectroscopy (EDS) while reducing data collection time by using non-local means filtering. Non-local means filtering has been a successful image processing method for denoising images by replacing pixel intensities through a non-local smoothing kernel. This technique was recently advanced to operate on high-dimensional datasets for the denoising of electron backscatter diffraction patterns (EBSPs) within an EBSD scan, which are denoised by averaging an EBSP with patterns within a certain distance that are determined to be alike by a similarity metric. This work examines using similar high-dimensional non-local means denoising to hyperspectral EDS data from nickel aluminum bronze, comparing data collected at longer dwell times to data collected at shorter dwell times with non-local means filtering applied. This enables an enhancement in spectral data quality with reduced collection time without worsening the spatial resolution. |
Proceedings Inclusion? |
Planned: |
Keywords |
Characterization, Computational Materials Science & Engineering, |