Denoising camera data - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 8

Table 2
Denoising Results for the Noise Level ASA 800

ASA800 Debayering PSNR PSNRHVS SSIM VIF

Proposed ADA 40.390 36.766 0.993 0.777

Proposed [ 15 ] 40.703 36.777 0.993 0.810

ClipFoi ADA 38.997 34.552 0.991 0.745

ClipFoi [ 15 ] 39.312 34.449 0.991 0.765

Zhang ADA 40.815 37.183 0.994 0.775

Zhang [ 15 ] 41.325 37.265 0.994 0.802

Table 3
Denoising Results for the Noise Level ASA 1600

ASA1600 Debayering PSNR PSNRHVS SSIM VIF

Proposed ADA 38.395 34.537 0.988 0.696

Proposed [ 15 ] 38.682 34.787 0.988 0.737

ClipFoi ADA 40.649 37.058 0.995 0.792

ClipFoi [ 15 ] 41.233 36.794 0.994 0.818

Zhang ADA 38.957 34.821 0.990 0.676

Zhang [ 15 ] 39.458 35.089 0.990 0.703

Table 4
Denoising Results for the Noise Level ASA 3200

ASA3200 Debayering PSNR PSNRHVS SSIM VIF

Proposed ADA 35.890 31.781 0.976 0.586

Proposed [ 15 ] 36.120 32.239 0.979 0.631

ClipFoi ADA 39.272 35.458 0.991 0.726

ClipFoi [ 15 ] 39.690 35.237 0.991 0.752

Zhang ADA 36.342 31.678 0.980 0.546

Zhang [ 15 ] 36.726 32.031 0.981 0.567

When comparing the results visually, however, we think our method provides slightly bet-
ter results. As the visual perception is not well represented by the metrics, we can evaluate