Spatial-Adaptive Network for Single Image Denoising

By Meng Chang, Qi Li, Huajun Feng, Zhi-hai Xu. 2020

190 citations recorded by the source corpus.

Previous works have shown that convolutional neural networks can achieve good performance in image denoising tasks. However, limited by the local rigid convolutional operation, these methods lead to oversmoothing artifacts. A deeper network structure could alleviate these problems, but more computational overhead is needed. In this paper, we propose a novel spatial-adaptive denoising network (SADNet) for efficient single image blind noise removal. To adapt to changes in spatial textures and edges, we design a residual spatial-adaptive block. Deformable convolution is introduced to sample the spatially correlated features for weighting. An encoder-decoder structure with a context block is introduced to capture multiscale information. With noise removal from the coarse to fine, a high-quality noisefree image can be obtained. We apply our method to both synthetic and real noisy image datasets. The experimental results demonstrate that our method can surpass the state-of-the-art denoising methods both quantitatively and visually.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Fahad Shahbaz Khan
2. Ming-Hsuan Yang
3. Ling Shao
4. Salman Hameed Khan
5. Luc Van Gool
6. Munawar Hayat
7. Syed Waqas Zamir
8. Aditya Arora

In Media Technology

1. Ming-Hsuan Yang
2. Salman Hameed Khan
3. Fahad Shahbaz Khan
4. Munawar Hayat
5. Syed Waqas Zamir
6. Aditya Arora
7. Ling Shao
8. Chong Mou

Citations in the corpus

• Deep Generalized Unfolding Networks for Image Restoration (2022, result): “We compare our DGUNet with several recent methods [10, 53, 76, 77, 79] and report the evaluation results (PSNR and SSIM) in Tab.”
• Restormer: Efficient Transformer for High-Resolution Image Restoration (2021, unknown): “Method DnCNN BM3D CBDNet* RIDNet* AINDNet* VDN SADNet* DANet+* CycleISP* MIRNet DeamNet* MPRNet DAGL Uformer Restormer Dataset [101] [15] [25] [6] [35] [89] [12] [90] [91] [92] [63] [93] [55] [80] (Ours)”
• Learning Enriched Features for Fast Image Restoration and Enhancement (2022, methodology): “Noisy RIDNet [33] AINDNet [42] SADNet [39] CycleISP [38] DANet [40] MIRNet-v2 CycleISP [38] AINDNet [42] DANet [40] SADNet [39] MIRNet-v2 (cols. 1, 3, 4) and in turn causes loss of details near dark and high-light areas.”
• Multi-Stage Progressive Image Restoration (2021, background): “Noisy Image CycleISP [86] AINDNet [40] DANet [85] SADNet [11] MPRNet (Ours)”