基于特征细化与注意力增强重构的水下图像增强算法

万刚,王小波,石纲,叶德震,朱思思,司帆

Underwater image enhancement algorithm based on feature refinement and attention-augmented reconstruction

Gang WAN,Xiaobo WANG,Gang SHI,Dezhen YE,Sisi ZHU,Fan SI

表 1 各算法在LSUI数据集中的图像质量客观评价指标

Tab.1 Objective evaluation metrics of image quality for various algorithms on LSUI

算法 MSE↓ PSNR↑ SSIM↑ Entropy UIQM↑ UCIQE↑ 原始图像 1609.682 17.491 0.864 6.887 4.117 0.542 HE[7] 2558.828 14.931 0.751 7.141 4.308 0.604 CLAHE[8] 1502.201 17.008 0.843 7.208 4.337 0.565 DEFOG[6] 2511.267 15.608 0.869 6.925 4.325 0.553 Ucolor[25] 352.193 22.911 0.898 7.234 4.389 0.562 URST[26] 394.476 22.705 0.905 7.321 4.177 0.585 U-Shape[18] 334.992 24.293 0.881 7.284 4.418 0.571 WFFP[27] 968.713 18.271 0.735 6.792 4.201 0.601 Spectroformer[28] 774.806 19.388 0.640 6.314 3.9596 0.552 本研究算法 253.558 25.421 0.893 7.488 4.461 0.592