Research map / Video / Video segmentation and tracking
Joint Adaptive Sparsity and Low-Rankness on the Fly: An Online Tensor Reconstruction Scheme for Video Denoising
Bihan Wen, Yanjun Li, Luke Pfister, Yoram Bresler
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 low-rank regularization to the adaptive sparse model (BM-TL to SALT) (ablation) | SALT · TUT and ASU, all noise levels | PSNR (dB): 0.6 dB average improvement of SALT over BM-TL | Improved |
| add adaptive sparse approximation to the low-rank model (BM-LR to SALT) (ablation) | SALT · TUT and ASU, all noise levels | PSNR (dB): 0.2 dB average improvement of SALT over BM-LR | Improved |
| replace fixed 3D DCT with online learned sparsifying transform (ablation) | BM-TL · TUT and ASU, all noise levels | PSNR (dB): 3.1 dB average improvement of SALT over BM-DCT | Improved |
| learned transform vs fixed DCT in sparse model, TUT sigma=20 (ablation) | BM-TL · TUT dataset, noise sigma=20 | PSNR (dB): 31.15 -> 32.84 | Improved |
| SALT vs VBM4D, averaged over 26 test videos | SALT · TUT and ASU, all noise levels | PSNR (dB): 1.3 dB improvement over VBM4D | Improved |
| SALT vs VBM3D, averaged over 26 test videos | SALT · TUT and ASU, all noise levels | PSNR (dB): 1.2 dB improvement over VBM3D | Improved |
| SALT vs VIDOSAT, averaged over 26 test videos | SALT · TUT and ASU, all noise levels | PSNR (dB): 1.0 dB improvement over VIDOSAT | Improved |
| SALT vs sKSVD, averaged over 26 test videos | SALT · TUT and ASU, all noise levels | PSNR (dB): 1.6 dB improvement over sKSVD | Improved |
| SALT vs fBM3D, averaged over 26 test videos | SALT · TUT and ASU, all noise levels | PSNR (dB): 3.6 dB improvement over fBM3D | Improved |
| SALT vs VIDOSAT on Gbicycle frame, sigma=20 | SALT · Gbicycle (TUT), sigma=20 | PSNR (dB): 31.80 -> 35.67 | Improved |
| SALT vs VBM4D on Stefan frame, sigma=20 | SALT · Stefan (ASU), sigma=20 | PSNR (dB): 28.56 -> 29.69 | Improved |
Closest papers by meaning
- Bayesian Low-Tubal-Rank Robust Tensor Factorization with Multi-Rank Determination (TPAMI 2019 · similarity 0.87)
- Multispectral Images Denoising by Intrinsic Tensor Sparsity Regularization (CVPR 2016 · similarity 0.87)
- Fully Trainable and Interpretable Non-Local Sparse Models for Image Restoration (ECCV 2020 · similarity 0.87)
- Hyper-Laplacian Regularized Unidirectional Low-Rank Tensor Recovery for Multispectral Image Denoising (CVPR 2017 · similarity 0.87)
- Deep Rank-One Tensor Functional Factorization for Multi-Dimensional Data Recovery (AAAI 2025 · similarity 0.86)
- Denoising as Adaptation: Noise-Space Domain Adaptation for Image Restoration (ICLR 2025 · similarity 0.85)
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.
