Table 1.
DAGAN | DAWGAN | DAWGAN-GP | SARA-GAN | |||||
---|---|---|---|---|---|---|---|---|
Mask: 1D Gaussian | PSNR | SSIM | PSNR | SSIM | PSNR | SSIM | PSNR | SSIM |
Sample rate: 10% | 34.0484 ± 4.81 | 0.9538 ± 2.20e-2 | 35.4455 ± 4.57 | 0.9654 ± 1.71e-2 | 36.3499 ± 4.27 | 0.9718 ± 1.37e-2 | 36.3926 ± 4.73 | 0.9713 ± 1.46e-2 |
Sample rate: 20% | 40.2395 ± 4.23 | 0.9857 ± 6.41e-3 | 41.4284 ± 4.45 | 0.9894 ± 5.18e-3 | 42.6101 ± 3.92 | 0.9923 ± 3.77e-3 | 43.2054 ± 4.49 | 0.9929 ± 3.71e-3 |
Sample rate: 30% | 40.8891 ± 4.46 | 0.9873 ± 5.88e-3 | 42.1798 ± 4.61 | 0.9907 ± 4.53e-3 | 42.9149 ± 3.69 | 0.9928 ± 3.41e-3 | 43.3522 ± 4.34 | 0.9931 ± 3.47e-3 |
Mask: 2D Gaussian | PSNR | SSIM | PSNR | SSIM | PSNR | SSIM | PSNR | SSIM |
Sample rate: 10% | 39.7242 ± 5.01 | 0.9801 ± 1.06e-2 | 40.7623 ± 4.68 | 0.9861 ± 7.75e-3 | 41.1885 ± 4.43 | 0.9876 ± 6.85e-3 | 41.6323 ± 5.25 | 0.9881 ± 6.83e-3 |
Sample rate: 20% | 41.5595 ± 4.93 | 0.9857 ± 7.74e-3 | 41.7733 ± 5.15 | 0.9880 ± 6.84e-3 | 42.9742 ± 4.61 | 0.9912 ± 5.05e-3 | 43.4991 ± 5.15 | 0.9920 ± 4.84e-3 |
Sample rate: 30% | 44.3886 ± 5.01 | 0.9934 ± 3.95e-3 | 44.2812 ± 5.11 | 0.9932 ± 4.03e-3 | 44.7868 ± 3.86 | 0.9947 ± 2.97e-3 | 45.7536 ± 4.99 | 0.9951 ± 3.03e-3 |
The bold value means that the experimental result value is the best.