Research Article
An Efficient Universal Noise Removal Algorithm Combining Spatial Gradient and Impulse Statistic
Table 4
Comparative restoration results in PSNR (dB) and MAE (the second row) for Gaussian noise.
| Method | Lena image | Boats image | | | | | | |
| Noisy image | 28.13 | 22.14 | 16.37 | 28.13 | 22.17 | 16.40 | 4.002 | 7.961 | 15.50 | 3.983 | 7.914 | 15.23 | 3 × 3 median filter | 32.13 | 28.44 | 23.44 | 30.25 | 27.43 | 23.07 | 2.267 | 3.702 | 6.744 | 2.734 | 4.086 | 7.059 | 5 × 5 median filter | 30.44 | 28.94 | 25.95 | 27.67 | 26.72 | 24.63 | 2.280 | 3.083 | 4.787 | 3.213 | 3.906 | 5.502 | Gaussian filter | 33.14 | 30.12 | 27.08 | 32.04 | 28.83 | 25.86 | 2.216 | 2.856 | 4.459 | 2.885 | 3.475 | 5.133 | Bilateral filter | 33.90 | 30.67 | 27.19 | 32.69 | 29.22 | 25.94 | 1.975 | 2.785 | 4.293 | 2.322 | 3.358 | 5.113 | ROAD-TRIF | 33.97 | 30.57 | 27.11 | 32.92 | 29.23 | 25.93 | 1.847 | 2.811 | 4.239 | 2.180 | 3.359 | 5.111 | SG-BF | 33.68 | 30.68 | 27.32 | 32.44 | 29.19 | 26.03 | 2.004 | 2.764 | 4.117 | 2.322 | 3.391 | 5.023 |
|
|