Real Image Denoising with Feature Attention

ICCV 2019  ·  Saeed Anwar, Nick Barnes ·

Deep convolutional neural networks perform better on images containing spatially invariant noise (synthetic noise); however, their performance is limited on real-noisy photographs and requires multiple stage network modeling. To advance the practicability of denoising algorithms, this paper proposes a novel single-stage blind real image denoising network (RIDNet) by employing a modular architecture... We use a residual on the residual structure to ease the flow of low-frequency information and apply feature attention to exploit the channel dependencies. Furthermore, the evaluation in terms of quantitative metrics and visual quality on three synthetic and four real noisy datasets against 19 state-of-the-art algorithms demonstrate the superiority of our RIDNet. read more

PDF Abstract ICCV 2019 PDF ICCV 2019 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Color Image Denoising BSD68 sigma15 RIDNet PSNR 34.01 # 1
Grayscale Image Denoising BSD68 sigma15 RIDNet PSNR 31.81 # 4
Grayscale Image Denoising BSD68 sigma25 RIDNet PSNR 29.34 # 4
Color Image Denoising BSD68 sigma25 RIDNet PSNR 31.37 # 1
Grayscale Image Denoising BSD68 sigma50 RIDNet PSNR 26.4 # 4
Color Image Denoising CBSD68 sigma50 RIDNet PSNR 28.14 # 2
Color Image Denoising Darmstadt Noise Dataset RIDNet (blind) PSNR (sRGB) 39.23 # 3
SSIM (sRGB) 0.9526 # 3
Image Denoising DND RIDNet PSNR (sRGB) 39.26 # 11
SSIM (sRGB) 0.953 # 8
Image Denoising SIDD RIDNet PSNR (sRGB) 38.71 # 11
SSIM (sRGB) 0.951 # 11

Methods


No methods listed for this paper. Add relevant methods here