Super-Resolution Reconstruction of License Plate Image Based on Gradual Back-Projection Network

Dianwei Wang; Yuanjie Hao; Ying Liu; Yongjun Xie; Haijun Song

doi:10.3788/LOP57.161002

Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161002(2020)

Super-Resolution Reconstruction of License Plate Image Based on Gradual Back-Projection Network

Dianwei Wang¹, Yuanjie Hao^1、*, Ying Liu¹, Yongjun Xie², and Haijun Song²

¹School of Telecommunication and Information Engineering, Xi'an University of Posts & Telecommunications, Xi'an, Shaanxi 710121, China;

²Xi'an Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Xi'an, Shaanxi 710119, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Flow chart of the our algorithm

Download full size

Fig. 2. Detection and extraction results of license plate area

Download full size

Fig. 3. Structure of GBPN

Download full size

Fig. 4. Structure of gradual back-projection unit. (a) Gradual up-projection unit; (b) gradual down-projection unit

Download full size

Fig. 5. Part license plate images in the test set

Download full size

Fig. 6. L1 losses and R_PSN of different network models. (a)-(b) DBPN, GBPN-1, GBPN-11; (c)-(d) DBPN, GBPN-2, GBPN-21

Download full size

Fig. 7. Subjective results of different super-resolution algorithms when the amplification factor is 4. (a) LR images; (b) bicubic interpolation algorithm; (c) DCSCN algorithm; (d) DBPN algorithm; (e) DBPN-GAN algorithm; (f) GBPN-11 algorithm; (g) ground images

Download full size

Table 1. Parameters of different network structures
View table
Table 1. Parameters of different network structures
Network DBPN GBPN-1 GBPN-11 GBPN-2 GBPN-21
Parameter 599566 682906 524410 536086 486970

Table 2. Average R_PSN of the test set when the amplification factor is 4unit: dB
View table
Table 2. Average R_PSN of the test set when the amplification factor is 4unit: dB
License plate Bicubic DCSCN DBPN DBPN-GAN GBPN-11 GBPN-21
Normal light 29.00 29.42 29.81 29.72 29.87 29.83
Low light 34.33 35.13 35.69 35.66 35.73 35.67

Table 3. Average S_SIM of the test set when the amplification factor is 4
View table
Table 3. Average S_SIM of the test set when the amplification factor is 4
License plate Bicubic DCSCN DBPN DBPN-GAN GBPN-11 GBPN-21
Normal light 0.8017 0.8329 0.8775 0.8698 0.8803 0.8781
Low light 0.9059 0.9273 0.9574 0.9464 0.9600 0.9584

Table 4. Reconstruction times of the two algorithmsunit: ms
View table
Table 4. Reconstruction times of the two algorithmsunit: ms
Algorithm Original algorithm Our algorithm
Image super-resolution License plate detection License plate super-resolution Total
Normal light 12.52 2.10 5.81 7.91
Low light 12.48 2.08 5.25 7.33

Tools

Get Citation

Copy Citation Text

Dianwei Wang, Yuanjie Hao, Ying Liu, Yongjun Xie, Haijun Song. Super-Resolution Reconstruction of License Plate Image Based on Gradual Back-Projection Network[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites