Compression Artifacts Reduction by a Deep Convolutional Network

Lossy compression introduces complex compression artifacts, particularly the blocking artifacts, ringing effects and blurring. Existing algorithms either focus on removing blocking artifacts and produce blurred output, or restores sharpened images that are accompanied with ringing effects. Inspired by the deep convolutional networks (DCN) on super-resolution, we formulate a compact and efficient network for seamless attenuation of different compression artifacts. We also demonstrate that a deeper model can be effectively trained with the features learned in a shallow network. Following a similar "easy to hard" idea, we systematically investigate several practical transfer settings and show the effectiveness of transfer learning in low-level vision problems. Our method shows superior performance than the state-of-the-arts both on the benchmark datasets and the real-world use case (i.e. Twitter). In addition, we show that our method can be applied as pre-processing to facilitate other low-level vision routines when they take compressed images as input.

PDF Abstract ICCV 2015 PDF ICCV 2015 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
JPEG Artifact Correction ICB (Quality 10 Color) ARCNN PSNR 30.06 # 6
PSNR-B 31.21 # 5
SSIM 0.779 # 5
JPEG Artifact Correction ICB (Quality 10 Grayscale) ARCNN PSNR 31.13 # 5
PSNR-B 30.97 # 5
SSIM 0.794 # 5
JPEG Artifact Correction ICB (Quality 20 Color) ARCNN PSNR 32.24 # 6
PSNR-B 32.53 # 6
SSIM 0.778 # 6
JPEG Artifact Correction ICB (Quality 20 Grayscale) ARCNN PSNR 35.04 # 4
PSNR-B 32.72 # 5
SSIM 0.905 # 3
JPEG Artifact Correction ICB (Quality 30 Color) ARCNN PSNR 33.31 # 4
PSNR-B 33.72 # 4
SSIM 0.807 # 4
JPEG Artifact Correction LIVE1 (Quality 10 Color) ARCNN PSNR 26.91 # 8
PSNR-B 26.92 # 8
SSIM 0.795 # 8
JPEG Artifact Correction Live1 (Quality 10 Grayscale) ARCNN PSNR 29.11 # 11
PSNR-B 29.07 # 8
SSIM 0.8235 # 8
JPEG Artifact Correction LIVE1 (Quality 20 Color) ARCNN PSNR 29.23 # 8
PSNR-B 29.24 # 8
SSIM 0.865 # 8
JPEG Artifact Correction LIVE1 (Quality 20 Grayscale) ARCNN PSNR 31.29 # 11
PSNR-B 31.37 # 7
SSIM 0.8891 # 6

Methods


No methods listed for this paper. Add relevant methods here