Polarization Image Interpolation Algorithm via Tensor Non-Negative Sparse Factorization

Junchao Zhang; Jianlai Chen; Haibo Luo; Degui Yang; Buge Liang

doi:10.3788/AOS202141.1411001

Journals >Acta Optica Sinica >Volume 41 >Issue 14 >Page 1411001 > Article

Acta Optica Sinica
Vol. 41, Issue 14, 1411001 (2021)

Polarization Image Interpolation Algorithm via Tensor Non-Negative Sparse Factorization

Junchao Zhang¹, Jianlai Chen^1、*, Haibo Luo², Degui Yang¹, and Buge Liang¹

Author Affiliations

¹School of Aeronautics and Astronautics, Central South University, Changsha, Hunan 410083, China

²Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang, Liaoning 110169, China

show less

DOI: 10.3788/AOS202141.1411001 Cite this Article Set citation alerts

Junchao Zhang, Jianlai Chen, Haibo Luo, Degui Yang, Buge Liang. Polarization Image Interpolation Algorithm via Tensor Non-Negative Sparse Factorization[J]. Acta Optica Sinica, 2021, 41(14): 1411001 Copy Citation Text

show less

Micropolarization system and polarization tensor data structure. (a) Arrangement pattern of micropolarizer; (b) schematic of tensor decomposition of polarization data

Fig. 1. Micropolarization system and polarization tensor data structure. (a) Arrangement pattern of micropolarizer; (b) schematic of tensor decomposition of polarization data

Download full size

Fig. 2. Interpolation flow of polarizing image in focal plane

Download full size

Fig. 3. Results of reconstruction of different algorithms in scenario 1. (a) Ref. [16]; (b) Ref. [17]; (c) Ref. [19]; (d) Ref. [20]; (e) proposed algorithm; (f) ground truth

Download full size

Fig. 4. Results of reconstruction of different algorithms in scenario 2. (a) Ref. [16]; (b) Ref. [17]; (c) Ref. [19]; (d) Ref. [20]; (e) proposed algorithm; (f) ground truth

Download full size

Fig. 5. Reconstruction results under different image block sizes in scenario 1. (a) 2×2; (b) 4×4; (c) 8×8; (d) ground truth

Download full size

Fig. 6. Reconstruction results under different image block sizes in scenario 2. (a) 2×2; (b) 4×4; (c) 8×8; (d) ground truth

Download full size

Fig. 7. Reconstruction results of different algorithms in measured data. (a) Ref. [16]; (b) Ref. [17]; (c) Ref. [19]; (d) Ref. [20]; (e) proposed algorithm

Download full size

Image	Ref.[16]	Ref.[17]	Ref.[19]	Ref.[20]	Proposedalgorithm
I₀	0.9172	0.9130	1.0161	0.8343	0.8289
I₄₅	0.9417	0.9331	1.0818	0.8586	0.8492
I₉₀	1.0278	1.0207	1.0431	0.9386	0.9234
I₁₃₅	1.0154	1.0065	1.1373	0.9214	0.9011
S₀	1.3521	1.3431	1.3120	1.2923	1.3955
DoLP	0.0052	0.0052	0.0068	0.0049	0.0041

Table 1. RMSE results of different algorithms

Image	Ref.[16]	Ref.[17]	Ref.[19]	Ref.[20]	Proposedalgorithm
I₀	49.1474	49.1852	48.4587	49.9669	50.0212
I₄₅	49.0598	49.1380	47.9479	49.8136	49.9674
I₉₀	48.2762	48.3278	48.2223	49.0320	49.1974
I₁₃₅	48.3604	48.4241	47.4757	49.1414	49.3548
S₀	45.9172	45.9708	46.2035	46.2743	45.6678
DoLP	45.8336	45.9552	43.7292	46.6113	48.2496

Table 2. PSNR results of different algorithms

Image	Ref.[16]	Ref.[17]	Ref.[19]	Ref.[20]	Proposedalgorithm
I₀	0.9817	0.9816	0.9793	0.9850	0.9844
I₄₅	0.9822	0.9822	0.9787	0.9856	0.9849
I₉₀	0.9805	0.9805	0.9799	0.9843	0.9837
I₁₃₅	0.9788	0.9787	0.9767	0.9826	0.9821
S₀	0.9856	0.9856	0.9841	0.9868	0.9843
DoLP	0.8815	0.8815	0.8776	0.9093	0.9335

Table 3. SSIM results of different algorithms

Algorithm	Ref. [16]	Ref. [17]	Ref. [19]	Ref. [20]	Proposed algorithm
Running time /s	0.48	0.26	0.22	308.00	278.00

Table 4. Running time of different algorithms

Junchao Zhang, Jianlai Chen, Haibo Luo, Degui Yang, Buge Liang. Polarization Image Interpolation Algorithm via Tensor Non-Negative Sparse Factorization[J]. Acta Optica Sinica, 2021, 41(14): 1411001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information