AI Paper Atlas

Research map / Image generation / Image restoration

A Physics-Based Noise Formation Model for Extreme Low-Light Raw Denoising

CVPR 2020 · Image generation › Image restoration · approach: Raw and low-light noise modeling · 242 citations (OpenAlex)

Check your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 changedWhereMeasuredEffect
Replace Gaussian-approximated shot noise with true Poisson shot noise (ablation)G+P noise model · SID SonyPSNR: 37.08 -> 38.31; 32.85 -> 34.39; 31.87 -> 33.37Helped
Use Tukey lambda rather than Gaussian read noise (ablation)G+P* noise model · SID SonyPSNR: 38.31 -> 39.10; 34.39 -> 36.46; 33.37 -> 35.69Helped
Add row noise (ablation)G*+P* noise model · SID SonyPSNR: 39.10 -> 39.23; 36.46 -> 36.89; 35.69 -> 36.01Helped
Add uniform quantization noise (ablation)G*+P*+R noise model · SID SonyPSNR: 39.23 -> 39.27; 36.89 -> 37.13; 36.01 -> 36.30Helped
Train with the full physics-based noise model instead of paired real dataU-Net · SID SonyPSNR: 38.60 -> 39.27; 37.08 -> 37.13; 36.29 -> 36.30Mixed
Train with the calibrated physics-based noise model instead of paired real dataU-Net · ELD camerasPSNRMixed
Use Gaussian read noise and Gaussian-approximated shot noise instead of the full model (ablation)U-Net · SID SonyPSNR: 39.27 -> 37.08; 37.13 -> 32.85; 36.30 -> 31.87Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy