Research map / Image generation / Image restoration
Computational Optics for Mobile Terminals in Mass Production
Shiqi Chen, Ting Lin, Huajun Feng, Zhihai Xu, Qi Li, Yueting Chen
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 |
|---|---|---|---|
| construct proxy camera from measured SFR instead of ideal design | proxy camera · Phone FoV 0.9 | SSIM: 0.806 -> 0.859 | Improved |
| optimize SFR area instead of pixel-level PSF MSE (sun et al.) | proxy camera · Phone FoV 0.9 | SFR MSE (x1e-3): 1.157 -> 0.952 | Improved |
| use proposed framework instead of CODEV tolerancing | proxy camera · DSLR FoV 0.9 | SFR MSE (x1e-3): 1.242 -> 0.877 | Improved |
| CODEV tolerancing median bias vs ideal design | CODEV · DSLR FoV 0.9 | SSIM: 0.872 -> 0.833 | Hurt |
| add dilated omni-dimensional dynamic convolution vs FDN, trained on multiple samples | DOConv model · DSLR multiple samples | PSNR: 41.94 -> 42.12 | Improved |
| add dilated omni-dimensional dynamic convolution vs FDN, trained on multiple samples | DOConv model · Phone multiple samples | PSNR: 31.86 -> 32.28 | Improved |
| dynamic model vs FDN when trained on single sample | DOConv model · DSLR single sample | PSNR: 43.12 -> 43.03 | No clear effect |
| dynamic model vs FDN when trained on single sample | DOConv model · Phone single sample | PSNR: 34.56 -> 34.28 | No clear effect |
| add imaging simulation to MTF calculation (vs continuous wavefront) (ablation) | proxy camera construction · proxy camera accuracy | SSIM | Improved |
| use SFR area instead of MTF at 0.5 Nyquist (ablation) | proxy camera construction · proxy camera accuracy | SSIM | Improved |
| use dynamic damping instead of fixed damping factor (ablation) | proxy camera construction · proxy camera accuracy | SSIM | Improved |
| replace ordinary convolution with dynamic convolution (ablation) | restoration model · multi-sample synthetic data | restoration quality | Improved |
Closest papers by meaning
- The Differentiable Lens: Compound Lens Search Over Glass Surfaces and Materials for Object Detection (CVPR 2023 · similarity 0.84)
- Towards Universal Computational Aberration Correction in Photographic Cameras: A Comprehensive Benchmark Analysis (CVPR 2026 · similarity 0.83)
- Extreme-Quality Computational Imaging via Degradation Framework (ICCV 2021 · similarity 0.82)
- Auto-Tuning Structured Light by Optical Stochastic Gradient Descent (CVPR 2020 · similarity 0.81)
- Oscillation-Reduced MXFP4 Training for Vision Transformers (ICML 2025 · similarity 0.80)
- A Physics-informed Low-rank Deep Neural Network for Blind and Universal Lens Aberration Correction (CVPR 2024 · 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.
