Distortion Correction Method of Interference Projection Based on Convolutional Neural Network

Meng Yan; Qitai Huang; Jianfeng Ren

doi:10.3788/LOP230636

Laser & Optoelectronics Progress, Volume. 61, Issue 8, 0805001(2024)

Distortion Correction Method of Interference Projection Based on Convolutional Neural Network

Meng Yan^1,2, Qitai Huang^1,2、*, and Jianfeng Ren^1,2

¹School of Optoelectronic Science and Engineering & Collaborative Innovation Center of Suzhou Nano Science and Technology, Soochow University, Suzhou 215006, Jiangsu, China

²Key Laboratory of Advanced Optical Manufacturing Technologies of Jiangsu Province & Key Laboratory of Modern Optical Technologies of Education Ministry of China, Soochow University, Suzhou 215006, Jiangsu, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(13)

Fig. 1. Manifestation of projection distortion. (a1)(a2) Radial distortion; (b1)(b2) decentering distortion; (c1)(c2) thin prism distortion; (d1)(d2) distortion superposition diagram

Download full size

Fig. 2. Simulation optical path of aspheric surface compensation detection

Download full size

Fig. 3. Partial images of the data set

Download full size

Fig. 4. Network structure of DCE Net model

Download full size

Fig. 5. Projection distortion correction based on CNN

Download full size

Fig. 6. RMSE of coordinate correction error obtained from 100 random experiments

Download full size

Fig. 7. Experimental picture. (a) Physical picture of compensation test; (b) interferogram (input layer)

Download full size

Fig. 8. Results of aspheric surface error detection. (a) Before correction; (b) after correction

Download full size

Table 1. Aspheric surface parameters to be tested
View table
Table 1. Aspheric surface parameters to be tested
Effective aperture /mm Radius of curvature /mm Conic 4^th order term 6^th order term 8^th order term 10^th order term
102.5 276.41 4.88 3.151×10^-8 1.343×10^-12 5.962×10^-17 5.514×10^-21

Table 2. Distortion coefficient
View table
Table 2. Distortion coefficient
No. k₁ k₂
A₁ -0.0481 0.0499
A₂ -0.0269 0.0448
A₃ -0.0417 0.0470

Table 3. Experimental results

View table

Table 3. Experimental results

Item	$δ$ /%	RMSE	Elapsed time
Net-MP	6.910	0.0085	226 min 1 s
Net-Without2AP	26.580	0.0313	825 min 18 s
Net-5×5	3.390	0.0059	464 min 27 s
AlexNet	5.105	0.0127	239 min 48 s
VGG11Net	4.735	0.0262	289 min 44 s
DCE Net	0.430	0.0047	206 min 30 s

Table 4. Distortion of feature points before and after distortion correction
View table
Table 4. Distortion of feature points before and after distortion correction
Feature point Before correction After correction
（92，63） 6.4546 1.0555
（112，78） 7.9949 0.8279
（156，108） 9.1243 1.2871

Table 5. Comparison of correction results of different methods
View table
Table 5. Comparison of correction results of different methods
Method of correction Position error RMSE /pixel
x'-x y'-y $φ$
Marker^［4］ 2.9482 2.1261 3.6349
Rapid correction^［17］ 1.2498 1.4661 1.9265
DCE Net 1.4514 1.0467 1.7895

Tools

Get Citation

Copy Citation Text

Meng Yan, Qitai Huang, Jianfeng Ren. Distortion Correction Method of Interference Projection Based on Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2024, 61(8): 0805001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites