Multi-polarization fusion generative adversarial networks for clear underwater imaging

Ding, Xueyan; Fu, Xianping

doi:10.1016/j.optlaseng.2022.106971

Cited by 29 publications

(13 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Polarization, one of the most important characteristics of electromagnetic waves, has been widely applied in many fields such as wireless communications, [1][2][3][4] biosensing, [5][6][7] and polarization imaging. [8][9][10][11] Especially, linearly polarized light is normally required for polarized light therapy, [12] head-up display system, [13] and LCD/OLED applications. [14] Hence, it is highly demanded to design and fabricate high performance polarization converters which could covert the natural light to one particular linearly polarized light with high efficiency to use the natural light more efficiently.…”

Section: Introductionmentioning

confidence: 99%

High efficiency and high transmission asymmetric polarization converter with chiral metasurface in visible and near-infrared region

GAO

Tian

et al. 2023

Chinese Phys. B

View full text Add to dashboard Cite

Polarization manipulation of light is of great importance which could promote the development of wireless communications, biosensing and polarization imaging. In order to use natural light more efficiently, it is highly demanded to design and fabricate high performance asymmetric polarization converters which could covert the natural light to one particular linearly polarized light with high efficiency. Traditionally, polarizers could be achieved by controllers with crystals and polymers exhibiting birefringence. However, the polarizers are bulky in size and the theoretical conversion efficiency of the polarizers is limited to 0.5 with unpolarized light incidence. In this paper, we propose a polarization converter which could preserve high transmission for one linearly polarized light and convert the orthogonal linearly polarized light to its cross-polarized with high transmittance based on a multi-layer chiral metasurface. Theoretical results show that normally incident $y$-polarized light preserves high transmittance for the wavelength range from 685 to 800 nm while the orthogonal normally incident $x$-polarized light is efficiently converted to the $y$-polarized light with high transmittance from 725 to 748 nm. Accordingly, for unpolarized light incidence, transmittance larger than 0.5 has been successfully achieved in a broadband wavelength range from 712 to 773 nm with a maximum transmittance of 0.58 at 732 nm.

show abstract

Section: Introductionmentioning

confidence: 99%

High efficiency and high transmission asymmetric polarization converter with chiral metasurface in visible and near-infrared region

GAO

Tian

et al. 2023

Chinese Phys. B

View full text Add to dashboard Cite

show abstract

“…In the past decade, machine learning and neural network have achieved a variety of successes in many areas including distance prediction [13], image dehazing [14], [15], visible light positioning (VLP) [16], and image recognition [17]. The deep learning method, as a sub-field of machine learning, has been recognized as an effective approach that leads to many technological advancements [18].…”

Section: Introductionmentioning

confidence: 99%

Underwater Polarization Imaging Recovery Based on Polarimetric Residual Dense Network

et al. 2022

View full text Add to dashboard Cite

Application of deep-learning to polarization imaging technology for image restoration has led to many technological breakthroughs, especially in underwater image recovery and recognition. In this work, a four-input deep learning model with the Polarimetric Residual Dense Network is proposed for underwater image recovery. The diverse polarization component images are trained and tested in different processes in the network for the recognition and dehazing by considering the physical model of polarization dehazing imaging. Our study reveals that the proposed method can efficiently recover the hazed images, and provide good performance for improving the quality of image restoration even in a high-turbidity complex underwater environment.

show abstract

“…Recently, the deep learning method has developed rapidly and has been considered a successful way to outperform the traditional intensity-based ones and boost performance in polarimetric imaging techniques, including denoising, demosaicing, and de-scattering tasks (Sun et al, 2021;Liu et al, 2022;Ding et al, 2022). In contrast to the physical model, deep learning methods don't require complex physical models and prior knowledge due to powerful fitting capability.…”

Section: Introductionmentioning

confidence: 99%

“…It can effectively remove the scattering light and obviously be more robust than other traditional methods; but this method cannot deal with color images. A medium scale of color polarization image datasets from natural conditions through passive polarization imaging was built by Ding (Ding et al, 2022) et al They constructed generative feature-fusion adversarial networks to extract different polarization angles features and obtain better results on both laboratory simulated and real natural datasets. Since passive polarization imaging relies on ambient light for illumination, the method is not suitable for the task of strong scattering environment de-scattering.…”

Section: Introductionmentioning

confidence: 99%

UCRNet: Underwater color image restoration via a polarization-guided convolutional neural network

Huang

et al. 2022

Front. Mar. Sci.

View full text Add to dashboard Cite

Underwater images always suffer from low contrast and color distortion due to the wavelength-dependent scattering and absorption effects caused by particles existing in turbid water, especially in high turbidity conditions. Based on the polarization properties of the backscattering light, polarimetric methods can estimate the intensity level of the backscattering and the transmittance of the media. Accordingly, they can separate the target signal from the undesired ones to achieve high-quality imaging. In addition, learning-based polarimetric methods are effective for gray-model image restoration, but the learning-based polarimetric technique for color image restoration has yet to be considered. In this paper, we propose a 3- dimensional convolutional neural network, which maintains the correlation of polarization information among different polarization channel images as well as embodies polarization constraints, for underwater color image restoration. The experimental results verify that the proposed solution improves the image quality (i.e., the image contrast, details, and color) and outperforms other existing methods, especially when the turbidity of scattering media is high. The proposed solution can be readily applied to practical applications and potentially realize the clear vision in other scattering media, including biomedical imaging and remote sensing.

show abstract

Multi-polarization fusion generative adversarial networks for clear underwater imaging

Cited by 29 publications

References 25 publications

High efficiency and high transmission asymmetric polarization converter with chiral metasurface in visible and near-infrared region

High efficiency and high transmission asymmetric polarization converter with chiral metasurface in visible and near-infrared region

Underwater Polarization Imaging Recovery Based on Polarimetric Residual Dense Network

UCRNet: Underwater color image restoration via a polarization-guided convolutional neural network

Contact Info

Product

Resources

About