Public

EBSD — Public Tier

(3 scenes)

Full-access development tier with all data visible.

Measurements (y), ideal forward operator (H), spec ranges, ground truth (x_true), and true mismatch spec.

Load HDF5 → compare reconstruction vs x_true → check consistency → iterate.

Reconstructed signals (x_hat) and corrected spec as HDF5.

✓

True spec visible — use these exact values for Scenario III oracle reconstruction.

Parameter	Spec Range	True Value	Unit
pattern_center	-2.0 – 4.0	1.0	pixels
sample_tilt	69.5 – 71.0	70.25	deg
detector_distance	-0.5 – 1.0	0.25	mm

Method: CPU_baseline — Mismatch parameter: nominal

14.02 dB

SSIM 0.0139

14.01 dB

SSIM 0.0134

17.95 dB

SSIM 0.5073

Per-scene breakdown (4 scenes)

Scene	PSNR I	SSIM I	PSNR II	SSIM II	PSNR III	SSIM III
scene_00	13.36	0.0125	13.36	0.0119	17.66	0.5450
scene_01	14.60	0.0131	14.60	0.0126	17.67	0.3923
scene_02	14.75	0.0176	14.75	0.0168	18.78	0.5438
scene_03	13.36	0.0127	13.36	0.0122	17.69	0.5481

#	Method	Score	PSNR	SSIM	Consistency	Trust	Source
1	DiffEBSD + gradient	0.835	36.11	0.974	0.92	✓ Certified	Gao et al., NeurIPS 2024
2	PhysEBSD + gradient	0.819	35.0	0.968	0.91	✓ Certified	Chen et al., Acta Mater. 2024
3	SwinEBSD + gradient	0.803	33.87	0.96	0.9	✓ Certified	Li et al., npj Comput. Mater. 2023
4	TransEBSD + gradient	0.780	32.36	0.946	0.89	✓ Certified	Wang et al., Acta Mater. 2022
5	PointEBSD + gradient	0.767	30.92	0.93	0.93	✓ Certified	Foden et al., Ultramicroscopy 2022
6	DnCNN-EBSD + gradient	0.721	27.94	0.88	0.95	✓ Certified	Kaufmann et al., npj Comput. Mater. 2020
7	TV-EBSD + gradient	0.637	24.56	0.788	0.88	✓ Certified	Wilkinson et al., Mater. Charact. 2006
8	DI-EBSD + gradient	0.578	22.32	0.703	0.87	✓ Certified	Chen et al., Ultramicroscopy 2015
9	Hough-EBSD + gradient	0.537	20.46	0.621	0.92	✓ Certified	Krieger Lassen, J. Microsc. 1994

y H_ideal spec_ranges x_true true_spec

Format: HDF5

Scenes: 3

0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

PSNR: 40% SSIM: 40% Consistency: 20%