Correction of Orbital Angular Momentum State Based on Diffractive Deep Neural Network

Kansong Chen; Bailin Liu; Chenghao Han; Shengmei Zhao; Le Wang; Haichao Zhan

doi:10.3788/AOS230663

Acta Optica Sinica, Volume. 43, Issue 24, 2401008(2023)

Correction of Orbital Angular Momentum State Based on Diffractive Deep Neural Network

Kansong Chen¹, Bailin Liu¹, Chenghao Han¹, Shengmei Zhao^1,2、*, Le Wang¹, and Haichao Zhan¹

¹School of Communications and Information Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210023, Jiangsu, China

²National Laboratory of Solid State Microstructures, Nanjing 210008, Jiangsu, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. Schematic diagram of random phase screen model for simulating turbulence

Download full size

Fig. 2. OAM correction model based on D2NN

Download full size

Fig. 3. Intensity maps of LG beam with different topological loads under different turbulent disturbances

Download full size

Fig. 4. Phase diagrams of LG beam with different topological loads under different turbulence interferences

Download full size

Fig. 5. Comparison before and after intensity and phase correction under different topological loads

Download full size

Fig. 6. PSNR value variation of OAM state intensity images with variation of iteration times of D2NN under different turbulence disturbances

Download full size

Fig. 7. OAM state correction effects under different topological loads

Download full size

Fig. 8. Trained loss function against epochs with different parameters

Download full size

Fig. 9. Tested loss function against epochs with different parameters

Download full size

Table 1. Intensity plots of LG beams and phase screens under different turbulence interferences
View table
Table 1. Intensity plots of LG beams and phase screens under different turbulence interferences
$C_{n}^{2}$ =10^-16 $C_{n}^{2}$ =10^-14 $C_{n}^{2}$ =10^-12
Intensity
Phase screen

Table 2. Loss function size and similarity magnitude under different network layers when only phase parameters are used in training
View table
Table 2. Loss function size and similarity magnitude under different network layers when only phase parameters are used in training
Index 5 layers 6 layers 7 layers 8 layers 9 layers
Train_loss 0.0012 0.0011 0.0007 0.0008 0.0009
Valid_loss 0.0010 0.0008 0.0007 0.0006 0.0008
SSIM 0.277 0.322 0.408 0.402 0.391

Table 3. Loss function size and similarity magnitude under different network layers when only amplitude parameters are used in training
View table
Table 3. Loss function size and similarity magnitude under different network layers when only amplitude parameters are used in training
Index 5 layers 6 layers 7 layers 8 layers 9 layers
Train_loss 0.0012 0.0008 0.0006 0.0006 0.0006
Valid_loss 0.0011 0.0009 0.0007 0.0006 0.0006
SSIM 0.521 0.667 0.714 0.802 0.801

Table 4. Loss function size and similarity magnitude under different network layers when phase and amplitude parameters are combined in training
View table
Table 4. Loss function size and similarity magnitude under different network layers when phase and amplitude parameters are combined in training
Index 5 layers 6 layers 7 layers 8 layers 9 layers
Train_loss 0.0013 0.0009 0.0006 0.0004 0.0004
Valid_loss 0.0011 0.0008 0.0007 0.0006 0.0006
SSIM 0.671 0.705 0.736 0.811 0.804

Tools

Get Citation

Copy Citation Text

Kansong Chen, Bailin Liu, Chenghao Han, Shengmei Zhao, Le Wang, Haichao Zhan. Correction of Orbital Angular Momentum State Based on Diffractive Deep Neural Network[J]. Acta Optica Sinica, 2023, 43(24): 2401008

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Atmospheric Optics and Oceanic Optics

Received: Mar. 13, 2023

Accepted: Jun. 5, 2023

Published Online: Dec. 12, 2023

The Author Email: Zhao Shengmei (zhaosm@njupt.edu.cn)

DOI:10.3788/AOS230663

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Intensity plots of LG beams and phase screens under different turbulence interferences

Table 1. Intensity plots of LG beams and phase screens under different turbulence interferences

Table 2. Loss function size and similarity magnitude under different network layers when only phase parameters are used in training

Table 2. Loss function size and similarity magnitude under different network layers when only phase parameters are used in training

Table 3. Loss function size and similarity magnitude under different network layers when only amplitude parameters are used in training

Table 3. Loss function size and similarity magnitude under different network layers when only amplitude parameters are used in training

Table 4. Loss function size and similarity magnitude under different network layers when phase and amplitude parameters are combined in training

Table 4. Loss function size and similarity magnitude under different network layers when phase and amplitude parameters are combined in training