Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
102 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Exploring Efficient Asymmetric Blind-Spots for Self-Supervised Denoising in Real-World Scenarios (2303.16783v2)

Published 29 Mar 2023 in cs.CV

Abstract: Self-supervised denoising has attracted widespread attention due to its ability to train without clean images. However, noise in real-world scenarios is often spatially correlated, which causes many self-supervised algorithms that assume pixel-wise independent noise to perform poorly. Recent works have attempted to break noise correlation with downsampling or neighborhood masking. However, denoising on downsampled subgraphs can lead to aliasing effects and loss of details due to a lower sampling rate. Furthermore, the neighborhood masking methods either come with high computational complexity or do not consider local spatial preservation during inference. Through the analysis of existing methods, we point out that the key to obtaining high-quality and texture-rich results in real-world self-supervised denoising tasks is to train at the original input resolution structure and use asymmetric operations during training and inference. Based on this, we propose Asymmetric Tunable Blind-Spot Network (AT-BSN), where the blind-spot size can be freely adjusted, thus better balancing noise correlation suppression and image local spatial destruction during training and inference. In addition, we regard the pre-trained AT-BSN as a meta-teacher network capable of generating various teacher networks by sampling different blind-spots. We propose a blind-spot based multi-teacher distillation strategy to distill a lightweight network, significantly improving performance. Experimental results on multiple datasets prove that our method achieves state-of-the-art, and is superior to other self-supervised algorithms in terms of computational overhead and visual effects.

PDF HTML Abstract
Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Bookmark Chat Bubble Oval Streamline Icon: https://streamlinehq.com Chat (Pro)
Definition Search Book Streamline Icon: https://streamlinehq.com
References (51)
  1. A high-quality denoising dataset for smartphone cameras. In CVPR, 2018.
  2. Real image denoising with feature attention. In ICCV, 2019.
  3. Noise2Self: Blind denoising by self-supervision. In ICML, 2019.
  4. Natural image noise dataset. In CVPR Workshops, 2019.
  5. GAN2GAN: Generative noise learning for blind image denoising with single noisy images. In ICLR, 2021.
  6. A. Chambolle. An algorithm for total variation minimization and applications. Journal of Mathematical Imaging and Vision, 20:89–97, 2004.
  7. Noise suppression in low-light images through joint denoising and demosaicing. In CVPR, 2011.
  8. Image blind denoising with generative adversarial network based noise modeling. In CVPR, 2018.
  9. Rethinking atrous convolution for semantic image segmentation. arXiv:1706.05587, 2017.
  10. NBNet: Noise basis learning for image denoising with subspace projection. In CVPR, 2021.
  11. Image denoising by sparse 3-D transform-domain collaborative filtering. IEEE TIP, 16(8):2080–2095, 2007.
  12. Image denoising via sparse and redundant representations over learned dictionaries. TIP, 15:3736–3745, 2006.
  13. srgb real noise synthesizing with neighboring correlation-aware noise model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1683–1691, 2023.
  14. Efficient knowledge distillation from an ensemble of teachers. In Interspeech, pages 3697–3701, 2017.
  15. Generative adversarial nets. In NIPS, 2014.
  16. Weighted nuclear norm minimization with application to image denoising. In CVPR, 2014.
  17. Toward convolutional blind denoising of real photographs. In CVPR, 2019.
  18. End-to-end unpaired image denoising with conditional adversarial networks. In AAAI, 2020.
  19. Pseudo 3D auto-correlation network for real image denoising. In CVPR, 2021.
  20. Neighbor2Neighbor: Self-supervised denoising from single noisy images. In CVPR, 2021.
  21. C2N: Practical generative noise modeling for real-world denoising. In ICCV, 2021.
  22. Self-supervised image denoising with downsampled invariance loss and conditional blind-spot network. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12196–12205, 2023.
  23. A review of an old dilemma: Demosaicking first, or denoising first? In CVPR Workshops, 2020.
  24. Transfer learning from synthetic to real-noise denoising with adaptive instance normalization. In CVPR, 2020.
  25. Noise2Void-learning denoising from single noisy images. In CVPR, 2019.
  26. High-quality self-supervised deep image denoising. In NeurIPS, 2019.
  27. Ap-bsn: Self-supervised denoising for real-world images via asymmetric pd and blind-spot network. In CVPR, pages 17725–17734, 2022.
  28. Noise2Noise: Learning image restoration without clean data. In ICML, 2018.
  29. Spatially adaptive self-supervised learning for real-world image denoising. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9914–9924, 2023.
  30. Multi-level Wavelet-CNN for image restoration. In CVPR Workshops, 2018.
  31. Invertible denoising network: A light solution for real noise removal. In CVPR, pages 13365–13374, 2021.
  32. Generative adaptive convolutions for real-world noisy image denoising. In AAAI, pages 1935–1943, 2022.
  33. Noisier2Noise: Learning to denoise from unpaired noisy data. In CVPR, 2020.
  34. Cvf-sid: Cyclic multi-variate function for self-supervised image denoising by disentangling noise from image. In CVPR, pages 17583–17591, 2022.
  35. Random sub-samples generation for self-supervised real image denoising. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12150–12159, 2023.
  36. Recorrupted-to-Recorrupted: Unsupervised deep learning for image denoising. In CVPR, 2021.
  37. A case for denoising before demosaicking color filter array data. In Asilomar Conference on Signals, Systems and Computers, 2009.
  38. Benchmarking denoising algorithms with real photographs. In CVPR, 2017.
  39. Modeling textures with total variation minimization and oscillating patterns in image processing. Journal of Scientific Computing, 19:553–572, 2003.
  40. Understanding convolution for semantic segmentation. In WACV, pages 1451–1460. Ieee, 2018.
  41. Blind2unblind: Self-supervised image denoising with visible blind spots. In CVPR, pages 2027–2036, 2022.
  42. Lg-bpn: Local and global blind-patch network for self-supervised real-world denoising. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18156–18165, 2023.
  43. Unpaired learning of deep image denoising. In ECCV, 2020.
  44. Noisy-As-Clean: Learning self-supervised denoising from corrupted image. IEEE TIP, 29:9316–9329, 2020.
  45. Multi-scale context aggregation by dilated convolutions. arXiv:1511.07122, 2015.
  46. Variational denoising network: Toward blind noise modeling and removal. In NeurIPS, 2019.
  47. Dual adversarial network: Toward real-world noise removal and noise generation. In ECCV, 2020.
  48. Beyond a Gaussian denoiser: Residual learning of deep CNN for image denoising. IEEE TIP, 26(7):3142–3155, 2017.
  49. FFDNet: Toward a fast and flexible solution for CNN-based image denoising. IEEE TIP, 27(9):4608–4622, 2018.
  50. Idr: Self-supervised image denoising via iterative data refinement. In CVPR, pages 2098–2107, 2022.
  51. When AWGN-based denoiser meets real noises. In AAAI, 2020.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (4)
  1. Shiyan Chen (6 papers)
  2. Jiyuan Zhang (57 papers)
  3. Zhaofei Yu (61 papers)
  4. Tiejun Huang (130 papers)
Citations (3)
View on Semantic Scholar