Dehazing and deblurring of underwater images with heavy-tailed priors

Shiwen Li; Shiwen Li; Feng Liu; Feng Liu; Jian Wei; Jian Wei

doi:10.1364/AO.452345

Applied Optics
Vol. 61,
Issue 13,
pp. 3855-3870
(2022)
•https://doi.org/10.1364/AO.452345

Dehazing and deblurring of underwater images with heavy-tailed priors

Shiwen Li, Feng Liu, and Jian Wei

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Shiwen Li, Feng Liu, and Jian Wei, "Dehazing and deblurring of underwater images with heavy-tailed priors," Appl. Opt. 61, 3855-3870 (2022)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Check for updates

Related Topics
Table of Contents Category
- Imaging Systems, Image Processing, and Displays
Optics & Photonics Topics
?

The topics in this list come from the Optics and Photonics Topics applied to this article.

About this Article
History
- Original Manuscript: December 28, 2021
- Revised Manuscript: March 20, 2022
- Manuscript Accepted: March 31, 2022
- Published: April 27, 2022

Abstract

The common problems of underwater images include color cast, haze effect, and the motion blur effect caused by turbulence and camera shake. To address these problems, research on color cast and haze and blur effects is carried out in this paper. Because red light has significant attenuation underwater, which could cause color cast of images, this paper proposes a red channel compensation method to solve this problem. This approach adaptively compensates for the red channel according to the pixel value of the red channel, successfully preventing excessive compensation. To address the haze effect of underwater images, combined with the physical model of underwater images, a variational method is introduced in the paper. This method can not only recover clear underwater images, but also refine the transmission map at the same time. Furthermore, the blind deconvolution method is adopted to deblur underwater images. First, the blur kernel of an underwater image is estimated, and then a clear underwater image is recovered based on the obtained blur kernel. Finally, qualitative and quantitative comparisons of the underwater images recovered by different methods are also carried out. From the qualitative perspective, the images recovered by our method have higher image sharpness and more outstanding details. The quantitative comparison results show that the images recovered using our method have higher scores according to various criteria. Therefore, on the whole, our method presents great advantages in comparison with others.

Full Article | PDF Article

More Like This

Underwater image restoration via adaptive color correction and dehazing

Jiening Zhang, Qing Yu, and Guojia Hou
Appl. Opt. 63(10) 2728-2736 (2024)

Fusion-based underwater image enhancement with category-specific color correction and dehazing

Yiming Li, Chunli Zhu, Junxin Peng, and Liheng Bian
Opt. Express 30(19) 33826-33841 (2022)

Underwater image enhancement using adaptive color restoration and dehazing

Tengyue Li, Shenghui Rong, Wenfeng Zhao, Long Chen, Yongbin Liu, Huiyu Zhou, and Bo He
Opt. Express 30(4) 6216-6235 (2022)

Previous Article Next Article

Supplementary Material (1)

Name	Description
Code 1	Underwater image processing code

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (11)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (11)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (28)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

	0.5	1	$e$	10
PCQI/UCIQE	1.0003/0.5815	1.0059/0.5817	1.0272/0.5888	1.0140/0.5841
FADE/UIQM	2.6662/1.6465	2.6608/1.6487	2.5352/1.7301	2.6436/1.6592

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 5(a)	0.6046	0.5177	0.5792	0.5327	0.5065	0.6565	0.5067	0.6537
Fig. 5(b)	0.5383	0.5620	0.6333	0.5659	0.5571	0.6430	0.5565	0.6736
Fig. 5(c)	0.5805	0.5594	0.6792	0.5380	0.5521	0.6231	0.5289	0.6799
Fig. 5(d)	0.5372	0.6363	0.5974	0.5560	0.5351	0.6715	0.6130	0.6861
Fig. 5(e)	0.5010	0.5731	0.5936	0.5453	0.5201	0.6113	0.5070	0.6161
Fig. 5(f)	0.5760	0.6541	0.5869	0.5476	0.5693	0.6308	0.5691	0.6600
Fig. 5(g)	0.5486	0.6093	0.5421	0.5272	0.5136	0.6675	0.5242	0.6026
Fig. 5(h)	0.5243	0.4697	0.6165	0.5593	0.5708	0.6493	0.5528	0.6538
Fig. 5(i)	0.5740	0.5210	0.6548	0.5613	0.5865	0.6498	0.5692	0.6948
Fig. 5(j)	0.5501	0.6236	0.6239	0.5388	0.5695	0.6292	0.5032	0.6267
Average	0.5536	0.5726	0.6106	0.5473	0.5481	0.6432	0.5431	0.6547

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 5(a)	0.4328	0.3275	0.5101	0.4786	0.6475	0.1743	0.7688	0.1362
Fig. 5(b)	0.2620	0.4729	0.3116	0.3475	0.2506	0.1942	0.3554	0.1419
Fig. 5(c)	0.3146	0.4700	0.3721	0.4348	0.3394	0.3115	0.4751	0.1874
Fig. 5(d)	0.4994	0.4730	0.5251	0.6571	0.4198	0.3886	0.3313	0.1667
Fig. 5(e)	0.4051	0.5313	0.3709	0.5063	0.4080	0.2471	0.6153	0.1436
Fig. 5(f)	0.5422	0.4089	0.8804	0.6282	0.7148	0.3999	0.7091	0.2864
Fig. 5(g)	0.5857	0.4211	0.5967	0.5844	0.4653	0.2580	0.7035	0.2142
Fig. 5(h)	0.4328	1.4563	0.5107	0.5644	0.4698	0.2263	0.6194	0.2521
Fig. 5(i)	0.2408	0.5827	0.2751	0.3281	0.2232	0.1884	0.2949	0.1147
Fig. 5(j)	0.5545	0.4788	0.5271	0.6094	0.4323	0.4123	0.8136	0.3253
Average	0.4269	0.5622	0.4879	0.5139	0.4371	0.2800	0.5686	0.1968

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 5(a)	3.0777	1.2537	2.8769	2.5738	3.1826	1.7102	2.7893	3.3680
Fig. 5(b)	2.8416	3.3706	3.3827	3.4589	3.4599	2.8361	3.5017	3.3909
Fig. 5(c)	2.5312	3.5194	3.3427	2.8729	3.4874	3.0490	3.1328	3.1680
Fig. 5(d)	2.9475	3.2187	2.9653	3.1906	3.4584	2.9283	3.5363	3.3100
Fig. 5(e)	3.0815	2.9977	3.2336	3.3204	3.3871	3.0049	3.3021	3.5199
Fig. 5(f)	2.6138	1.4771	2.8199	2.4493	2.8508	1.5877	2.6944	2.6693
Fig. 5(g)	2.9139	2.0818	2.9765	3.2125	3.4628	2.4407	3.1802	3.2179
Fig. 5(h)	2.7615	2.6227	3.1331	3.0705	3.3999	2.8426	3.1512	3.5137
Fig. 5(i)	2.5418	3.4780	3.2445	3.5057	3.5175	2.9919	3.5640	3.5481
Fig. 5(j)	2.7447	2.6512	2.6891	2.7662	2.8868	2.1479	2.7824	2.2292
Average	2.8055	2.6670	3.0664	3.0420	3.3093	2.5539	3.1634	3.1935

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 6(a)	0.5260	0.5432	0.6042	0.5846	0.5429	0.6576	0.5713	0.6225
Fig. 6(b)	0.5422	0.4436	0.5806	0.5912	0.5289	0.6033	0.5029	0.5829
Fig. 6(c)	0.5158	0.5166	0.6893	0.6167	0.5669	0.6287	0.5356	0.7059
Fig. 6(d)	0.5083	0.4627	0.6176	0.5745	0.4992	0.6095	0.5340	0.5955
Fig. 6(e)	0.5115	0.6117	0.6846	0.5976	0.5783	0.6434	0.5871	0.6740
Fig. 6(f)	0.5553	0.4835	0.5736	0.5430	0.5094	0.6121	0.5528	0.5890
Fig. 6(g)	0.5387	0.6942	0.6571	0.6105	0.5558	0.6684	0.6084	0.6599
Average	0.5282	0.5365	0.6295	0.5883	0.5402	0.6318	0.5560	0.6328

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 6(a)	0.3833	0.5912	0.8493	0.5444	0.5943	0.3546	0.6547	0.2602
Fig. 6(b)	0.5268	0.8826	0.7605	0.7030	0.7697	0.3689	0.9974	0.3677
Fig. 6(c)	0.3829	1.3146	0.7008	0.6894	0.7111	0.4214	0.7868	0.1830
Fig. 6(d)	0.7173	1.6774	1.6645	1.8001	1.7784	0.5073	1.0637	0.4986
Fig. 6(e)	0.3693	0.5017	0.4486	0.4265	0.3490	0.2893	0.4847	0.1747
Fig. 6(f)	0.7460	0.9586	0.9522	1.1657	0.5360	0.3333	0.7687	0.2377
Fig. 6(g)	0.3801	0.2663	0.3982	0.3872	0.3345	0.2175	0.4436	0.1421
Average	0.5008	0.8846	0.8248	0.8166	0.7247	0.3560	0.7428	0.2662

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
Fig. 6(a)	2.6192	2.8429	3.0389	3.0137	3.2410	2.5864	3.1089	3.4297
Fig. 6(b)	2.9307	2.6780	2.9054	2.8062	3.1402	2.7819	2.5937	3.3188
Fig. 6(c)	2.4880	2.5936	2.9085	2.8533	2.9623	2.7580	2.7452	3.4056
Fig. 6(d)	2.2447	1.8740	2.0539	2.0297	1.9271	2.4213	2.2204	2.7103
Fig. 6(e)	2.8322	3.1660	2.9192	3.0693	3.2755	2.5277	3.2698	3.2650
Fig. 6(f)	2.7406	2.8068	2.5978	2.8121	3.2293	3.1567	3.1015	3.2112
Fig. 6(g)	3.2187	2.7648	3.3483	3.4689	3.4415	2.9481	3.5172	3.6207
Average	2.7248	2.6751	2.8245	2.8647	3.0310	2.7400	2.9367	3.2801

	UIE [19]	Song’s [16]	RDCP [15]	Li’s [11]	ACDC [28]	DPATN [20]	Ucolor [18]	Ours
UIQM	3.0089	2.8852	3.0073	3.1487	3.0904	3.0020	3.1657	3.2534
UCIQE	0.5168	0.5324	0.6237	0.5558	0.5829	0.6245	0.5905	0.6339
FADE	0.3897	0.8382	0.4934	0.6504	0.4824	0.3375	0.6367	0.2917
PCQI	0.9112	1.1141	1.0832	1.1473	1.2037	1.2307	1.1199	1.1927

	$λ_{1} = β_{1} = γ_{1} = 0$	$λ_{1} + β_{1} + γ_{1}$	$α_{2} + δ$	$α_{2} + δ + β_{2}$
Fig. 9(a)	0.6167/0.5942	0.6140/0.6425	0.6183/0.4367	0.6224/0.3286
Fig. 9(b)	0.7268/0.2267	0.6921/0.3115	0.7093/0.2906	0.7245/0.1958
Fig. 9(c)	0.6592/0.2428	0.6653/0.3268	0.6843/0.2662	0.7116/0.1406

Abstract

Supplementary Material (1)

Data availability

Cited By

Figures (11)

Tables (11)

Equations (28)

Applied Optics