Expansion of Sub-Light Field Data Based on Neural Light Field

Fig. 3. EPI of two sub-light fields and an extended light field after camera array data converted into biplanar data. (a) Sub-light field $L_{1}$ ; (b) sub-light field $L_{2}$ ; (c) extended light field

Download full size

Fig. 4. Rays in extended light field corresponding to camera array data. (a) Before expansion;(b) viewpoint is in overlapping area; (c) viewpoint is in non-overlapping area

Download full size

Fig. 5. Overview of entire process

Download full size

Fig. 6. Structure of neural light field network

Download full size

Fig. 7. Sub-aperture images of view point (3,5). (a)‒(d) Predicted images; (e)‒(h) ground truth

Download full size

Fig. 8. Sub-aperture images of view point (3,10). (a)—(d) Predicted images; (e)—(h) ground truth

Download full size

Fig. 9. EPI of expanding light fields with different proportion of sub-light field overlapping rays in total rays of the sub-light field. (a) (e) (i) 42.9%; (b) (f) (j) 62.5%; (c) (g) (k) 77.8%; (d) (h) (l) ground truth

Download full size

Table 1. Quantitative evaluation of sub-regions of extended light field
View table
Table 1. Quantitative evaluation of sub-regions of extended light field
R MSE $↓$ PSNR $↑$ SSIM $↑$
$R_{2}$ 16.413 35.981 0.920
$R_{0}^{1}$ 25.584 34.102 0.879
$R_{0}^{2}$ 18.999 35.353 0.891
$R_{1}^{1}$ 16.442 35.974 0.920
$R_{0}^{4}$ 23.285 34.465 0.899
$R_{0}^{3}$ 40.794 32.180 0.708

Table 2. Quantitative evaluation of different regions of extended light field with different overlapping area sizes

View table

Table 2. Quantitative evaluation of different regions of extended light field with different overlapping area sizes

R	M/%	MSE $↓$	PSNR $↑$	SSIM $↑$
$R_{2}$	77.8	16.413	35.981	0.920
	62.5	17.837	35.632	0.900
	42.9	18.183	35.535	0.904
$R_{0}^{1}$	77.8	25.584	34.102	0.879
	62.5	27.753	33.754	0.850
	42.9	29.559	33.441	0.833
$R_{0}^{2}$	77.8	18.999	35.353	0.891
	62.5	21.639	34.803	0.855
	42.9	21.766	34.770	0.855
$R_{1}^{1}$	77.8	16.442	35.974	0.920
	62.5	19.041	35.3677	0.880
	42.9	20.471	35.019	0.868
$R_{0}^{4}$	77.8	23.285	34.465	0.899
	62.5	27.062	33.807	0.848
	42.9	29.382	33.457	0.841
$R_{0}^{3}$	77.8	40.794	32.180	0.708
	62.5	41.749	32.040	0.682
	42.9	45.312	31.689	0.644

Table 3. Impact of overlap region size on quality of light field expansion
View table
Table 3. Impact of overlap region size on quality of light field expansion
M N
42.9 82.91
62.5 83.55
77.8 84.68

Tools

Get Citation

Copy Citation Text

Yan Shen, Di He, Chang Liu, Jun Qiu. Expansion of Sub-Light Field Data Based on Neural Light Field[J]. Laser & Optoelectronics Progress, 2024, 61(24): 2437009

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites