Fig 1.
The enhancement effect of current low-light enhancement algorithms on underwater low-light images.
Fig 2.
Underwater image imaging model.
Fig 3.
Network structure diagram.
Fig 4.
The filling process of the bright channel image.
Fig 5.
Flow chart of feature extraction mechanism after adding hybrid attention module.
Fig 6.
Comparison experiment between the proposed method and the methods in this paper.
Table 1.
Image quality evaluation index of various methods on the same test set.
Fig 7.
Ablation experiments, the second column shows that the bright channel prior method is removed from the overall network; the third column shows that the hybrid attention module is removed after removing the prior method of bright channel; the fourth column indicates that on the overall network, only the feature retention loss is used.
Table 2.
Corresponding indexes of each method in the ablation experiment.
Fig 8.
Imbalance between brightness enhancement and deblurring in a few images.