Research map / Image generation / Image restoration
Burst Denoising via Temporally Shifted Wavelet Transforms
Xuejian Rong, Denis Demandolx, Kevin Matzen, Priyam Chatterjee, Yingli Tian
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 high-resolution feature encoding (HRNet-style) to U-Net baseline (ablation) | burst denoiser · DIV2K synthetic static bursts | PSNR | Improved |
| add explicit high-frequency (Haar wavelet pooling) features (ablation) | burst denoiser · DIV2K synthetic static bursts | PSNR | Improved |
| use temporal feature shifting instead of temporal max-pooling (ablation) | burst denoiser · DIV2K synthetic static bursts | PSNR | Improved |
| use temporal feature shifting instead of 3D convolution with high-res and high-freq features (ablation) | burst denoiser · DIV2K synthetic static bursts | PSNR | Improved |
| use temporal feature shifting instead of 3D convolution on baseline backbone (ablation) | burst denoiser · DIV2K synthetic static bursts | PSNR | No clear effect |
| replace KPN with the proposed model | burst denoiser · DIV2K synthetic static bursts | PSNR: 32.64 -> 32.88 | Improved |
| replace KPN with the proposed model | burst denoiser · DIV2K synthetic static bursts | SSIM: 0.8920 -> 0.8855 | Hurt |
| replace KPN with the proposed model | burst denoiser · Vimeo90K synthetic dynamic bursts | PSNR: 28.67 -> 28.61 | Hurt |
| replace KPN with the proposed model | burst denoiser · Vimeo90K synthetic dynamic bursts | LPIPS: 0.2357 -> 0.2366 | Hurt |
| replace KPN with the proposed model | burst denoiser · DIV2K synthetic static bursts | LPIPS: 0.1772 -> 0.1742 | Improved |
| replace KPN with the proposed model | burst denoiser · Latency per 720x720 frame | latency (s): 0.597 -> 0.064 | Improved |
| use burst averaging instead of single noisy reference frame | naive baseline · Vimeo90K synthetic dynamic bursts | PSNR: 23.33 -> 21.39 | Hurt |
Closest papers by meaning
- Deep Burst Denoising (ECCV 2018 · similarity 0.91)
- Iterative Residual CNNs for Burst Photography Applications (CVPR 2019 · similarity 0.89)
- Burst Denoising With Kernel Prediction Networks (CVPR 2018 · similarity 0.88)
- Efficient Burst Raw Denoising with Variance Stabilization and Multi-frequency Denoising Network (IJCV 2022 · similarity 0.88)
- Real-World Mobile Image Denoising Dataset with Efficient Baselines (CVPR 2024 · similarity 0.87)
- Burst Image Deblurring Using Permutation Invariant Convolutional Neural Networks (ECCV 2018 · similarity 0.87)
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.
