Polarization Image Denoising Based on Unsupervised Learning

Haofeng Hu; Huifeng Jin; Xiaobo Li; Jingsheng Zhai; Tiegen Liu

doi:10.3788/AOS221645

Acta Optica Sinica, Volume. 43, Issue 4, 0410001(2023)

Polarization Image Denoising Based on Unsupervised Learning

Haofeng Hu^1,2,3, Huifeng Jin^1,2, Xiaobo Li^3、*, Jingsheng Zhai³, and Tiegen Liu^1,2

Author Affiliations

¹School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China

²Key Laboratory of Optoelectronic Information Technology, Ministry of Education, Tianjin University, Tianjin 300072, China

³School of Marine Science and Technology, Tianjin University, Tianjin 300072, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Framework of unsupervised polarimetric-image denoising generative adversarial network

Download full size

Fig. 2. Structure of generative adversarial network. (a) Structure of generators; (b) structure of discriminators

Download full size

Fig. 3. Processing results of noisy images with different network structures

Download full size

Fig. 4. Processing results of indoor noise images by different methods

Download full size

Fig. 5. Denoising results of four different materials

Download full size

Fig. 6. Processing results of outdoor noise images by different methods

Download full size

Table 1. Indicator comparison of intensity, DoLP, and AoP images for different network structures

View table

Table 1. Indicator comparison of intensity, DoLP, and AoP images for different network structures

Network		Net-1	Net-2	Net-3	Net-4
D_DoLP		×	√	×	√
D_AoP		×	×	√	√
Intensity image	PSNR /dB	23.7057	26.3142	31.2859	31.3139
Intensity image	SSIM	0.7104	0.8874	0.8970	0.8877
DoLP image	PSNR /dB	19.2122	24.5586	26.2351	27.6377
DoLP image	SSIM	0.5517	0.6536	0.6631	0.6849
AoP image	PSNR /dB	13.3240	12.7801	13.9141	15.4566
AoP image	SSIM	0.1912	0.1912	0.1979	0.2123

Table 2. Indicator comparison of intensity, DoLP, and AoP images for different methods

View table

Table 2. Indicator comparison of intensity, DoLP, and AoP images for different methods

Method	Intensity image		DoLP image		AoP image
Method	SSIM	PSNR /dB	SSIM	PSNR /dB	SSIM	PSNR /dB
BM3D	0.8863	33.3374	0.2940	23.4488	0.1942	13.6127
PDRDN	0.8928	31.4548	0.6843	25.2243	0.2049	15.8199
Proposed	0.8877	31.3139	0.6849	27.6377	0.2123	15.4566

Table 3. Indicator comparison of intensity, DoLP, and AoP images for different materials

View table

Table 3. Indicator comparison of intensity, DoLP, and AoP images for different materials

Material	Intensity image		DoLP image		AoP image
Material	SSIM	PSNR /dB	SSIM	PSNR /dB	SSIM	PSNR /dB
Resin	0.9007	32.5596	0.6703	27.3557	0.2071	15.7295
Fabric	0.9207	33.4452	0.7776	30.1627	0.2462	15.8806
Wood	0.8755	31.0895	0.6748	30.2264	0.2319	16.6111
Plastic	0.9181	32.8964	0.7755	27.9045	0.2409	16.4649

Tools

Get Citation

Copy Citation Text

Haofeng Hu, Huifeng Jin, Xiaobo Li, Jingsheng Zhai, Tiegen Liu. Polarization Image Denoising Based on Unsupervised Learning[J]. Acta Optica Sinica, 2023, 43(4): 0410001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Aug. 26, 2022

Accepted: Sep. 22, 2022

Published Online: Feb. 22, 2023

The Author Email: Li Xiaobo (lixiaobo@tju.edu.cn)

DOI:10.3788/AOS221645

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology