Research map / Image generation / Image restoration
A Hybrid Frequency-Spatial Domain Model for Sparse Image Reconstruction in Scanning Transmission Electron Microscopy
Bintao He, Fa Zhang, Huanshui Zhang, Renmin Han
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add non-local patch enhancement (ablation) | FSHNet · Synthetic STEM, 5% sampling | PSNR (dB): 25.86 -> 26.72 | Improved |
| add non-local patch enhancement (ablation) | FSHNet · Synthetic STEM, 1.56% sampling | PSNR (dB): 19.97 -> 21.22 | Improved |
| add non-local patch enhancement (ablation) | FSHNet · Synthetic STEM, 9.375% sampling | PSNR (dB): 26.96 -> 26.80 | No clear effect |
| add global structure prior guidance (ablation) | FSHNet · Synthetic STEM, 5% sampling | PSNR (dB): 25.84 -> 26.72 | Improved |
| add frequency-domain initial reconstruction (ablation) | FSHNet · Synthetic STEM, 5% sampling | PSNR (dB): 22.07 -> 26.72 | Improved |
| add frequency-domain initial reconstruction (ablation) | FSHNet · Synthetic STEM, 1.56% sampling | PSNR (dB): 21.05 -> 21.22 | No clear effect |
| replace linear spatial initial reconstruction with frequency-domain one (ablation) | FSHNet · Synthetic STEM, 5% sampling | PSNR (dB): 22.07 -> 26.72 | Improved |
| replace CNN baselines with FSHNet | CSNet · Synthetic STEM, 1.56% sampling, dose 10 | PSNR (dB): 18.28 -> 25.22 | Improved |
| increase sampling ratio at test time for a model trained at 5% | FSHNet · Synthetic STEM robustness (train 5%) | PSNR (dB): 21.22 -> 26.80 (1.56% -> 9.375%) | Improved |
| test below the training sampling ratio (1.56% vs trained 5%) | FSHNet · Synthetic STEM robustness (train 5%) | PSNR (dB): 25.22 -> 21.22 | Hurt |
| increase electron dose per pixel from 10 to 25 | FSHNet · Synthetic STEM, 8% sampling | PSNR (dB): 26.34 -> 27.85 | Improved |
| use FSHNet instead of CSNet for speed | FSHNet · Runtime per image | seconds: 0.4824 -> 0.3788 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReconNet — ReconNet: Non-Iterative Reconstruction of Images From Compressively Sensed Measurements (CVPR 2016) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Spatial-Frequency Learning for Scanning Transmission Electron Microscopy Image Enhancement (TPAMI 2026 · similarity 0.93)
- Noise Calibration and Spatial-Frequency Interactive Network for STEM Image Enhancement (CVPR 2025 · similarity 0.90)
- Clean Implicit 3D Structure From Noisy 2D STEM Images (CVPR 2022 · similarity 0.88)
- Statistical Characteristic-Guided Denoising for Rapid High-Resolution Transmission Electron Microscopy Imaging (CVPR 2026 · similarity 0.88)
- Learning Large-Factor EM Image Super-Resolution with Generative Priors (CVPR 2024 · similarity 0.83)
- A Gaussian Filter-Based 3D Registration Method for Series Section Electron Microscopy (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
