Research map / Image generation / Image restoration
Making the Invisible Visible: Toward High-Quality Terahertz Tomographic Imaging via Physics-Guided Restoration
Weng-Tai Su, Yi-Chun Hung, Po-Jen Yu, Shang‐Hua Yang, Chia‐Wen Lin
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 subspace-attention fusion (SAFM) to amplitude-feature U-Net (ablation) | U-Net+Amp · THz restoration, Box | PSNR (dB): 20.32 -> 21.83 | Improved |
| add subspace-attention fusion (SAFM) to amplitude-feature U-Net (ablation) | U-Net+Amp · THz restoration, Skull | PSNR (dB): 20.70 -> 20.18 | Hurt |
| add SAFM to phase-feature U-Net (ablation) | U-Net+Phase · THz restoration, Box | PSNR (dB): 20.42 -> 21.65 | Improved |
| add SAFM when fusing amplitude and phase (ablation) | SARNet · THz restoration, Box | PSNR (dB): 20.00 -> 22.67 | Improved |
| add SAFM when fusing amplitude and phase (ablation) | SARNet · THz restoration, Skull | PSNR (dB): 21.18 -> 22.48 | Improved |
| add subspace projection to SARNet (ablation) | SARNet · THz restoration, Box | PSNR (dB): 21.43 -> 22.67 | Improved |
| add multi-view inter-view fusion to SARNet (ablation) | SARNet · THz restoration, Box | PSNR (dB): 22.67 -> 23.23 | Improved |
| add multi-view inter-view fusion to SARNet (ablation) | SARNet · THz restoration, Skull | PSNR (dB): 22.48 -> 23.05 | Improved |
| add multi-spectral amplitude features without SAFM to U-Net (ablation) | U-Net · THz restoration, Box | PSNR (dB): 19.77 -> 20.32 | Improved |
| combine amplitude and phase directly without SAFM instead of amplitude only (ablation) | SARNet w/o SAFM · THz restoration, Box | PSNR (dB): 20.32 -> 20.00 | Hurt |
| add SAFM to multi-view model (ablation) | SARNet_MV · THz restoration, Box | PSNR (dB): 22.10 -> 23.23 | Improved |
| use SARNet_MV instead of NBNet | SARNet_MV · THz restoration, average | PSNR (dB): +2.23 dB | 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.
- NBNet — NBNet: Noise Basis Learning for Image Denoising With Subspace Projection (CVPR 2021) · better on 9 of 9 shared results
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
- OpticalNet: An Optical Imaging Dataset and Benchmark Beyond the Diffraction Limit (CVPR 2025 · similarity 0.81)
- Single View Refractive Index Tomography with Neural Fields (CVPR 2024 · similarity 0.79)
- WaveMo: Learning Wavefront Modulations to See Through Scattering (CVPR 2024 · similarity 0.79)
- NTR-Gaussian: Nighttime Dynamic Thermal Reconstruction with 4D Gaussian Splatting Based on Thermodynamics (CVPR 2025 · similarity 0.79)
- V2V3D: View-to-View Denoised 3D Reconstruction for Light Field Microscopy (CVPR 2025 · similarity 0.79)
- TransmissiveGS: Residual-Guided Disentangled Gaussian Splatting for Transmissive Scene Reconstruction and Rendering (NeurIPS 2026 · similarity 0.79)
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.
