Research map / Image generation / Image restoration
Integrated Forward–Inverse Network for Reconstruction for Lensless Image Reconstruction
Donggeon Bae, Jaewoo Jung, Yong Guk Kang, Kyung Chul Lee, Taeyoung Kim, Jongho Kim, Sangjun Byun, Joonsik Park, Seung Ah Lee
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 forward and inverse operators (vs identity FSO/ISO) (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 24.674 -> 29.535 | Improved |
| add ISO to FSO-only (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 27.123 -> 29.535 | Improved |
| add FSO to ISO-only (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 28.711 -> 29.535 | Improved |
| use refinement block instead of plain convolutional gate (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 29.014 -> 29.535 | Improved |
| add learnable 2D regularizer (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 28.880 -> 29.535 | Improved |
| add initial ISO coarse estimate (ablation) | IFIN · DiffuserCam (k=1) | PSNR: 29.443 -> 29.535 | No clear effect |
| scale PSF field from k=1 to k=16 (ablation) | IFIN · DiffuserCam | PSNR: 29.535 -> 29.862 | Improved |
| learn PSF field instead of freezing calibrated PSFs (ablation) | IFIN · MultiWienerNet (k=9) | PSNR: 30.350 -> 31.083 | Improved |
| replace MWNet with IFIN | MWNet · MultiWienerNet | PSNR: 25.660 -> 31.083 | Improved |
| replace best prior method with IFIN | UPDN · DiffuserCam | PSNR: +1.63 dB | Improved |
| IFIN vs NAFNet at mild blur | NAFNet · Gaussian deblurring (sigma=5) | PSNR: 25.625 -> 25.100 | Hurt |
| IFIN vs NAFNet at strong blur | NAFNet · Gaussian deblurring (sigma=15) | PSNR: 21.535 -> 21.654 | Improved |
Closest papers by meaning
- FlatNet: Towards Photorealistic Scene Reconstruction from Lensless Measurements (TPAMI 2020 · similarity 0.89)
- Learning Physics-based Forward Model Corrections in Unrolled Networks for Diffuser-based Imaging (ECCV 2026 · similarity 0.88)
- Deep Non-Line-of-Sight Reconstruction (CVPR 2020 · similarity 0.85)
- Physics-Based Iterative Projection Complex Neural Network for Phase Retrieval in Lensless Microscopy Imaging (CVPR 2021 · similarity 0.85)
- PhoCoLens: Photorealistic and Consistent Reconstruction in Lensless Imaging (NeurIPS 2024 · similarity 0.84)
- LIRM: Large Inverse Rendering Model for Progressive Reconstruction of Shape, Materials and View-dependent Radiance Fields (CVPR 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.
