Enhancing microstructural fidelity in near-field high-energy diffraction microscopy reconstruction with Noise2Void denoising

Guo, X.; Beniwal, D.; McLain, D.; Almer, J.; Park, J.-S.; Kenesei, P.; Sharma, H.

doi:10.1107/S1600576726007612

Figure 7
Quantitative grain recovery against the EBSD ground truth. (a) Recall as a function of EBSD grain cross-sectional area for the N2V and traditional pipelines: both recover essentially all large grains, while N2V's advantage is concentrated in the small-to-intermediate range (50–400 µm²). (b) Decomposition of the EBSD grain population by outcome and size (recovered by both, by N2V only or by traditional only, or missed by both); the grains missed by both are overwhelmingly the smallest – the quantitative answer to why NF-HEDM recovers only a fraction of the EBSD grains. (c) Grain-detection probability versus size, following the analysis framework of Sparks et al. (2024

) (their Fig. 13): the fitted error function slopes reproduce the reported behaviour to ∼1%, and N2V shifts the curve toward smaller grains. Absolute diameters are 2D cross-sectional and so sit below the 3D volume-equivalent values of Sparks et al. (2024

JOURNAL OF
APPLIED
CRYSTALLOGRAPHY

ISSN: 1600-5767

Volume 59| Part 5| October 2026|

https://doi.org/10.1107/S1600576726007612

Open

access