Research map / Image generation / Image restoration
Unsupervised Non-Rigid Image Distortion Removal via Grid Deformation
Nianyi Li, Simron Thapa, Cameron Whyte, Albert W. Reed, Suren Jayasuriya, Jinwei Ye
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 iterative refinement after initialization (ablation) | Our init. · Simulated air turbulence (strong) | SSIM: 0.667 -> 0.799 | Improved |
| add iterative refinement after initialization (ablation) | Our init. · Simulated air turbulence (weak) | PSNR: 25.20 -> 24.29 | Hurt |
| use 4 conv layers per grid deformer instead of 2 (ablation) | grid deformer · Simulated air turbulence | PSNR: 19.23 -> 20.48 | Improved |
| use 4 conv layers instead of 6 (ablation) | grid deformer · Simulated air turbulence | PSNR: 20.06 -> 20.48 | Improved |
| per-frame grid deformers instead of one autoencoder predicting all grids (ablation) | grid deformer · Simulated air turbulence | PSNR: 16.83 -> 20.48 | Improved |
| increase number of input frames from 2 to 10 (ablation) | unsupervised network · Simulated air turbulence | PSNR: 17.55 -> 20.50 | Improved |
| increase number of input frames from 10 to 20 (ablation) | unsupervised network · Simulated air turbulence | PSNR: 20.50 -> 21.13 | No clear effect |
| increase number of input frames from 10 to 20 (running time) (ablation) | unsupervised network · Simulated air turbulence | running time: 265s -> 499s | Hurt |
| add Gaussian random Fourier feature position encoding to image generator (ablation) | image generator · Simulated air turbulence | SSIM: 30.9% improvement in SSIM and 13.9% improvement in PSNR | Improved |
| add initialization step for grid deformer and image generator (ablation) | unsupervised network · Simulated air turbulence | restoration quality | Improved |
| add initialization losses to the main optimization loop instead of separate step (ablation) | unsupervised network · Simulated air turbulence | restoration quality | Hurt |
| add regularizer R(I_k) (ablation) | unsupervised network · Simulated air turbulence | PSNR and SSIM | Improved |
Closest papers by meaning
- Looking Into the Water by Unsupervised Learning of the Surface Shape (NeurIPS 2025 · similarity 0.85)
- Learning Phase Distortion with Selective State Space Models for Video Turbulence Mitigation (CVPR 2025 · similarity 0.84)
- Turb-Seg-Res: A Segment-then-Restore Pipeline for Dynamic Videos with Atmospheric Turbulence (CVPR 2024 · similarity 0.84)
- NB-GTR: Narrow-Band Guided Turbulence Removal (CVPR 2024 · similarity 0.84)
- Diffeomorphic Template Registration for Atmospheric Turbulence Mitigation (CVPR 2024 · similarity 0.83)
- Decoupling Scattering: Pseudo-Label Guided NeRF for Scenes with Scattering Media (AAAI 2025 · similarity 0.83)
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.
