Laser & Optoelectronics Progress, Volume. 60, Issue 16, 1610007(2023)
Underwater Image Restoration Based on Total Variation and Color Balance
Figures & Tables(10)
Fig. 2. An example of estimating global background light. (a) Original image; (b) result of the quad-tree subdivision; (c) estimated global background light in red rectangle
Fig. 3. Comparison of the estimated transmission maps using different methods. (a) Original image; (b)-(d) the transmission maps obtained by DCP, UDCP, and the proposed method; (e)-(g) the corresponding restored results by DCP, UDCP, and the proposed method
Fig. 4. Restoration results of different algorithms. (a) Original images; (b) RCP algorithm; (c) IBLA algorithm; (d) ULAP algorithm;(e) UNTV algorithm; (f) UWCNN algorithm; (g) Bayes algorithm; (h) proposed algorithm
Fig. 5. Comparison of color correction. (a) Color checker; (b) original image; (c) RCP algorithm;(d) IBLA algorithm;(e) ULAP algorithm; (f) UNTV algorithm; (g) UWCNN algorithm; (h) Bayes algorithm; (i) proposed algorithm
Fig. 6. Ablation experiment. (a) Original images; (b) only defogging and deblurring; (c) only color correction; (d) proposed algorithm
Table 1. Comparison of UCIQE metrics of restored images with different algorithms
View tableTable 1. Comparison of UCIQE metrics of restored images with different algorithms
No. Raw RCP IBLA ULAP UNTV UWCNN Bayes Proposed algorithm image1 0.4559 0.5260 0.5673 0.6318 0.6182 0.4361 0.5023 0.6375 image2 0.5663 0.5783 0.6196 0.6502 0.6350 0.5067 0.5069 0.6690 image3 0.5008 0.5416 0.5865 0.5960 0.6297 0.4622 0.4909 0.7836 image4 0.4175 0.5337 0.4778 0.4944 0.5898 0.4531 0.4103 0.5971 image5 0.4526 0.5272 0.5366 0.5501 0.6204 0.4466 0.5101 0.6319 image6 0.5854 0.5725 0.6452 0.4788 0.6269 0.4744 0.4417 0.6465
Table 2. Comparison of UIQM metrics of restored images with different algorithms
View tableTable 2. Comparison of UIQM metrics of restored images with different algorithms
No. Raw RCP IBLA ULAP UNTV UWCNN Bayes Proposed algorithm image1 1.2010 1.4231 1.4409 1.6084 1.6436 1.1926 1.3137 1.6701 image2 1.1416 1.2280 1.3431 1.4736 1.5994 1.0312 0.9869 1.6736 image3 1.3831 1.4789 1.5752 1.5690 1.6937 1.3032 1.3174 1.7225 image4 1.4258 1.6014 1.5811 1.6896 1.6138 1.3671 1.2782 1.6931 image5 1.1904 1.5081 1.4057 1.4389 1.7019 1.2654 1.3186 1.7652 image6 1.7742 1.8249 1.8080 1.6488 1.8968 1.4976 0.9466 1.9033
Table 3. Comparison of FDUM metrics of restored images with different algorithms
View tableTable 3. Comparison of FDUM metrics of restored images with different algorithms
No. Raw RCP IBLA ULAP UNTV UWCNN Bayes Proposed algorithm image1 0.4613 0.5919 0.7031 0.9675 1.0101 0.3799 0.5250 1.0702 image2 0.5349 0.5946 0.6621 0.7610 1.0062 0.3469 0.3712 1.0318 image3 0.5152 0.6049 0.7320 0.7315 1.0499 0.4009 0.4735 1.0546 image4 0.3443 0.4723 0.5045 0.5266 0.7696 0.3603 0.2928 0.7730 image5 0.4596 0.6407 0.7216 0.7463 1.1097 0.4261 0.5874 1.1893 image6 0.5783 0.5768 0.8328 0.5263 0.9532 0.3336 0.3146 1.0874
Table 4. Comparison of average values of different objective quality evaluation metrics
View tableTable 4. Comparison of average values of different objective quality evaluation metrics
Metric Raw RCP IBLA ULAP UNTV UWCNN Bayes Proposed algorithm UCIQE 0.5121 0.5629 0.5969 0.6053 0.5978 0.5739 0.5184 0.6237 UIQM 1.2147 1.3995 1.4468 1.41 1.6147 1.3728 1.0467 1.6546 FDUM 0.4612 0.5393 0.6684 0.6575 0.8730 0.5552 0.5018 0.8845
Tools
Get Citation
Copy Citation Text
Yali Hao, Guojia Hou, Yuemei Li, Baoxiang Huang, Zhenkuan Pan. Underwater Image Restoration Based on Total Variation and Color Balance[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1610007
Paper Information
Category: Image Processing
Received: Sep. 1, 2022
Accepted: Oct. 18, 2022
Published Online: Aug. 15, 2023
The Author Email: Hou Guojia (hgj2015@qdu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20