Light-Field Image Quality Assessment Based on Multiple Visual Feature Aggregation

Zhuocheng Zou; Jun Qiu; Chang Liu

doi:10.3788/AOS202141.1610002

Acta Optica Sinica, Volume. 41, Issue 16, 1610002(2021)

Light-Field Image Quality Assessment Based on Multiple Visual Feature Aggregation

Zhuocheng Zou, Jun Qiu, and Chang Liu^*

Author Affiliations

Institute of Applied Mathematics, Beijing Information Science & Technology University, Beijing 100101, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. Framework of the proposed method. (a) Center view of the SAI; (b) macro-pixel image; (c) EPI; (d) RI sequence

Download full size

Fig. 2. Distribution curves of MSCN coefficients in different scenes. (a) Bikes; (b) Flowers

Download full size

Fig. 3. Macro-pixel images with different degradation level

Download full size

Fig. 4. Statistical distribution of two GLCM features on MLI with different degradation level. (a) Homogeneity; (b) entropy

Download full size

Fig. 5. Influence of different distortions on EPI. (a) Center view of light field SAI; (b) horizontal and vertical EPI with different degradation levels

Download full size

Fig. 6. Statistical distributions of three GLCM features on EPI. (a) Energy; (b) contrast; (c) homogeneity

Download full size

Fig. 7. Refocusing principle model

Download full size

Fig. 8. Refocused images on different depth layers

Download full size

Fig. 9. Statistical distribution of block entropy of refocused image at focus position a=1

Download full size

Fig. 10. Box plots of each index distribution in 1000 trials. (a) Results on SMART dataset; (b) results on Win5-LID dataset

Download full size

Table 1. Performance comparison on SMART dataset

View table

Table 1. Performance comparison on SMART dataset

Method	PLCC	SROCC	KROCC	RMSE
PSNR	0.6514	0.6883	0.4839	1.6366
SSIM	0.7180	0.6492	0.4623	1.5013
MS-SSIM	0.7578	0.6927	0.5498	1.4178
FSIM	0.7007	0.7398	0.5361	1.5389
VSI	0.7687	0.7726	0.5661	1.3796
VIF	0.5496	0.5078	0.3417	1.8020
VSNR	0.3659	0.4248	0.2798	2.0076
MDFM	0.7392	0.6468	0.4515	1.4528
LF-IQM	0.2998	0.1222	-	2.0579
LFG-LFC	0.8276	0.8246	0.6258	1.2108
Proposedmethod	0.8901	0.8711	0.7074	0.9246

Table 2. Performance comparison on Win5-LID Dataset

View table

Table 2. Performance comparison on Win5-LID Dataset

Method	PLCC	SROCC	RMSE
PSNR	0.6189	0.6026	0.8031
SSIM	0.7596	0.7346	0.6650
MS-SSIM	0.8388	0.8266	0.5566
FSIM	0.8318	0.8233	0.5675
VIF	0.7032	0.6665	0.7270
VSNR	0.5050	0.3961	0.8826
LF-IQM	0.4763	0.4503	0.8991
LF-QMLI	0.9038	0.8802	0.4147
Proposed method	0.9389	0.8698	0.3655

Table 3. Performance comparison of different features on SMART and Win5-LID datasets

View table

Table 3. Performance comparison of different features on SMART and Win5-LID datasets

Index	SMART				Win5-LID
Index	$\|F_{CV}\|$	$\|F_{MLI}\|$	$\|F_{EPI}\|$	$\|F_{RI}\|$	$\|F_{CV}\|$	$\|F_{MLI}\|$	$\|F_{EPI}\|$	$\|F_{RI}\|$
SROCC	0.6487	0.3596	0.7252	0.4913	0.7246	0.7776	0.7878	0.8189
KROCC	0.4766	0.2474	0.5415	0.3439	0.5522	0.5950	0.6040	0.6382
PLCC	0.7091	0.4495	0.7526	0.6712	0.7678	0.8008	0.8643	0.9030
RMSE	1.4633	1.8670	1.3958	1.5486	0.6727	0.6245	0.5254	0.4465

Tools

Get Citation

Copy Citation Text

Zhuocheng Zou, Jun Qiu, Chang Liu. Light-Field Image Quality Assessment Based on Multiple Visual Feature Aggregation[J]. Acta Optica Sinica, 2021, 41(16): 1610002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Jan. 28, 2021

Accepted: Mar. 18, 2021

Published Online: Aug. 12, 2021

The Author Email: Liu Chang (liuchang@bistu.edu.cn)

DOI:10.3788/AOS202141.1610002

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology