Spatial-Adaptive Network for Single Image Denoising

TL;DR

The paper introduces SADNet, a novel spatial-adaptive neural network that effectively removes noise from images by capturing multiscale information and adapting to local textures, outperforming existing methods.

Contribution

Proposes a residual spatial-adaptive block with deformable convolution and an encoder-decoder structure for efficient blind image denoising.

Findings

Outperforms state-of-the-art denoising methods quantitatively.

Achieves superior visual quality in denoised images.

Effective on both synthetic and real noisy datasets.

Abstract

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.