Table 4 Ablation study results showing the effect of outlier exclusion and soft thresholding on denoising metrics at varying noise levels.
Noise level (σ) | Method | PSNR | SSIM | FSIM | Entropy | BRISQUE | NIQE | PIQE |
|---|---|---|---|---|---|---|---|---|
10 | Proposed | 34.97 | 0.94 | 0.95 | 6.7 | 43.22 | 5.13 | 56.44 |
10 | No merging | 33.7 | 0.91 | 0.93 | 6.65 | 46.85 | 5.45 | 58.21 |
10 | No soft thresholding | 34.1 | 0.92 | 0.94 | 6.68 | 45.32 | 5.28 | 57.49 |
20 | Proposed | 31.79 | 0.9 | 0.92 | 6.69 | 46.04 | 4.78 | 61.16 |
20 | No merging | 30.55 | 0.87 | 0.9 | 6.63 | 48.7 | 5.05 | 62.89 |
20 | No soft thresholding | 30.92 | 0.88 | 0.91 | 6.66 | 47.29 | 4.95 | 62.1 |
30 | Proposed | 29.8 | 0.86 | 0.9 | 6.73 | 44.28 | 4.69 | 56.74 |
30 | No merging | 28.52 | 0.83 | 0.88 | 6.68 | 46.92 | 4.92 | 58.1 |
30 | No soft thresholding | 28.95 | 0.84 | 0.89 | 6.71 | 45.88 | 4.79 | 57.53 |
40 | Proposed | 28.38 | 0.82 | 0.87 | 6.87 | 43.58 | 5.13 | 54.41 |
40 | No merging | 27.05 | 0.79 | 0.85 | 6.8 | 46.11 | 5.29 | 55.96 |
40 | No soft thresholding | 27.53 | 0.8 | 0.86 | 6.84 | 44.96 | 5.2 | 55.12 |
