Research Article
Global Distribution Adjustment and Nonlinear Feature Transformation for Automatic Colorization
Table 1
The results of learning process before and after applying M-step.
| Input | Color correlation | E-step | M-step | Feature correlation | PSNR | Feature correlation | PSNR |
| | 4.65218 | 1.70593 | 26.6034 | 2.98258 | 26.8436 | | 6.17421 | 1.85361 | 27.4339 | 3.54563 | 27.6762 | | 6.52125 | 1.97957 | 25.3198 | 5.1082 | 25.5997 | | 5.45406 | 1.80969 | 26.3328 | 3.33898 | 27.0535 | | 2.63201 | 1.59904 | 27.2102 | 2.61883 | 27.5409 | | 4.21849 | 1.71946 | 27.0152 | 4.97776 | 27.6046 | | 8.3623 | 1.49676 | 26.1107 | 2.59245 | 26.3373 | | 3.81443 | 1.9547 | 23.039 | 3.78879 | 23.6149 | | 6.04486 | 2.06075 | 23.9355 | 5.5012 | 25.3075 | | 8.65296 | 2.36828 | 27.2904 | 6.48897 | 27.6309 | | 6.20452 | 2.36094 | 23.3158 | 5.9459 | 23.5803 | | 5.69085 | 2.15709 | 27.2325 | 5.66991 | 27.804 |
|
|