Research Article
CNN-Enabled Visibility Enhancement Framework for Vessel Detection under Haze Environment
Table 3
NIQE comparison of various image enhancement methods on all test images shown in Figure
6. The best results are highlighted in bold.
| | DCP | GRM | HL | F-LDCP | MSCNN | AOD-Net | GCA-Net | Ours |
| Image 7 | 5.448 | 5.780 | 5.296 | 5.932 | 6.224 | 7.220 | 5.080 | 5.057 | Image 8 | 5.525 | 10.297 | 4.961 | 5.868 | 5.826 | 5.716 | 5.694 | 5.122 | Image 9 | 4.856 | 5.281 | 5.204 | 5.330 | 5.255 | 5.409 | 4.626 | 5.103 | Average | 5.276 ± 0.299 | 7.119 ± 2.299 | 5.154 ± 0.141 | 5.710 ± 0.270 | 5.768 ± 0.398 | 6.115 ± 0.792 | 5.133 ± 0.438 | 5.094 ± 0.027 |
|
|