RCL Reconstruction Algorithm Based on Dynamically Updated Hybrid <i>a Priori</i> Constraints

Xinxin Lin; Zhiting Chen; Chuandong Tan; Liming Duan

doi:10.3788/AOS250531

Acta Optica Sinica, Volume. 45, Issue 11, 1134002(2025)

RCL Reconstruction Algorithm Based on Dynamically Updated Hybrid a Priori Constraints

Xinxin Lin^1,2, Zhiting Chen¹, Chuandong Tan¹, and Liming Duan^1、*

¹ICT Research Center, Key Laboratory of Optoelectronic Technology and Systems, Ministry of Education, Chongqing University, Chongqing 400044, China

²College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(22)

Fig. 1. Schematic diagrams of RCL imaging system. (a) RCL system structure; (b) RCL system geometric structure

Download full size

Fig. 2. Flowchart of DUHP algorithm

Download full size

Fig. 3. Process of extracting and updating a priori information on self-structures

Download full size

Fig. 4. Images of simulation models. (a) 3D view of model 1; (b) 3D view of model 2

Download full size

Fig. 5. Reconstruction results of simulation model 1 at the same noise level. The second to sixth columns show the reconstructed slices using SIRT, SART-TV, SART-TAwTV, SPI-TV, and DUHP. (a)‒(r) Slices of 4th, 14th, and 24th layers in the xoy plane; (s)‒(x) 100th layer slice in the xoz plane

Download full size

Fig. 6. ROIs of reconstruction detail area for simulation model 1. The second to sixth columns show zoomed-in views of reconstructed detail regions using SIRT, SART-TV, SART-TAwTV, SPI-TV, and DUHP. (a)‒(f) Rectangular ROI1 in the 14th layer of the xoy plane; (g)‒(l) rectangular ROI2 in the 100th layer of the xoz plane

Download full size

Fig. 7. Grayscale profile and error analysis along the dashed line in the 4th layer of xoy plane for different algorithms. (a) Grayscale distribution; (b) normalized error analysis

Download full size

Fig. 8. RMSE convergence and SSP of comparison algorithms at the same noise level. (a) RMSE convergence curves of comparison algorithms; (b) SSP along the solid line in the 100th layer of xoz plane for comparison algorithms

Download full size

Fig. 9. Reconstruction results of simulation model 2 at the same noise level. The second to sixth columns represent reconstruction slices of SIRT, SART-TV, SART-TAwTV, SPI-TV, and DUHP. (a1)‒(r1) Slices of 6th, 17th, and 27th layers in the xoy plane; (a2)‒(r2) difference maps of these slices; (s1)‒(x1) slices of 300th layer in the xoz plane; (s2)‒(x2) difference maps of 300th layer slices in the xoz plane

Download full size

Fig. 10. Grayscale profile and error analysis along the dashed line in the 17th layer of xoy plane for different algorithms. (a) Grayscale distribution; (b) normalized error analysis

Download full size

Fig. 11. RMSE convergence and SSP of comparison algorithms at the same noise level. (a) RMSE convergence curves of comparison algorithms; (b) SSP of comparison algorithms along the solid line in the 300th layer of xoz plane

Download full size

Fig. 12. Experimental system and its test object. (a) RCL scanning system; (b) decoder module circuit board sample; (c) projection data

Download full size

Fig. 13. Reconstruction results of decoder module circuit board. The first to fifth columns show the reconstructed slices using SIRT, SART-TV, SART-TAwTV, SPI-TV, and DUHP. (a)‒(e) 3D reconstruction results of the decoder module circuit board; (f)‒(t) slices of 12th, 17th, and 19th layers in the xoy plane; (u)‒(y) 217th layer slice in the xoz plane

Download full size

Fig. 14. RMSE convergence curves for different self-prior weights

Download full size

Fig. 15. RMSE convergence curves under different a priori information introduction timings $t_{0}$

Download full size

Table 1. Simulation experiment scanning parameters
View table
Table 1. Simulation experiment scanning parameters
Parameter Value
R_CL /mm 200
D_CL /mm 50
Tilt angle $θ$ /（°） 45
Pixel number of detector 1024×1024
Detector pixel size /（mm×mm） 0.1×0.1
Number of projections 64
Scanning method Equal angle

Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms
View table
Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms
Algorithm SIRT SART-TV SART-TAwTV SPI-TV DUHP
RMSE 0.1276 0.0536 0.0396 0.0170 0.0093
SSIM 0.6512 0.9298 0.9763 0.9909 0.9987
PSNR 17.88 25.42 28.06 35.41 40.59

Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms
View table
Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms
Algorithm SIRT SART-TV SART-TAwTV SPI-TV DUHP
RMSE 0.2705 0.1719 0.1631 0.0677 0.0301
SSIM 0.4190 0.6779 0.7585 0.9665 0.9890
PSNR 11.36 15.29 15.75 23.38 30.41

Table 4. Geometric parameters of RCL system

View table

Table 4. Geometric parameters of RCL system

Parameter	Value
R_CL /mm	606.7
D_CL /mm	152.9
Tilt angle $θ$ /（°）	50
Pixel number of detector	1536×1536
Detector pixel size /（mm×mm）	0.3×0.3
Number of projections	128
Scanning method	Equal angle
Reconstruction scale	150/70

Table 5. Algorithm computational performance comparison
View table
Table 5. Algorithm computational performance comparison
Algorithm Iterationtime /s GPU memoryusage /MB Projectiondata /MB
SIRT 1.70 1233.70 256
SART-TV 13.80 1253.57 256
SART-TAwTV 13.70 1248.57 256
SPI-TV 29.15 1225.01 256
DUHP 20.33 1264.07 256

Table 6. Quantitative evaluation parameters with different self-prior weights $α$
View table
Table 6. Quantitative evaluation parameters with different self-prior weights $α$
$α$ $0.05$ $0.20$ $0.50$ $0.60$ $0.80$
RMSE 0.0118 0.0105 0.0093 0.0113 0.0115
SSIM 0.9964 0.9978 0.9987 0.9985 0.9985
PSNR 36.67 38.57 40.59 38.96 38.77

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀
View table
Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀
t₀ 10 20 30 40 50
RMSE 0.0291 0.0226 0.0093 0.0122 0.0157
SSIM 0.9937 0.9967 0.9987 0.9984 0.9980
PSNR 30.71 32.90 40.59 38.26 36.06

Tools

Get Citation

Copy Citation Text

Xinxin Lin, Zhiting Chen, Chuandong Tan, Liming Duan. RCL Reconstruction Algorithm Based on Dynamically Updated Hybrid a Priori Constraints[J]. Acta Optica Sinica, 2025, 45(11): 1134002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: X-Ray Optics

Received: Jan. 21, 2025

Accepted: Apr. 22, 2025

Published Online: Jun. 24, 2025

The Author Email: Liming Duan (duanliming@cqu.edu.cn)

DOI:10.3788/AOS250531

CSTR:32393.14.AOS250531

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Simulation experiment scanning parameters

Table 1. Simulation experiment scanning parameters

Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms

Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms

Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms

Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms

Table 4. Geometric parameters of RCL system

Table 4. Geometric parameters of RCL system

Table 5. Algorithm computational performance comparison

Table 5. Algorithm computational performance comparison

Table 6. Quantitative evaluation parameters with different self-prior weights $α$

Table 6. Quantitative evaluation parameters with different self-prior weights $α$

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀

Table 1. Simulation experiment scanning parameters

Table 1. Simulation experiment scanning parameters

Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms

Table 2. Quantitative evaluation parameters of model 1 reconstruction results of different algorithms

Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms

Table 3. Quantitative evaluation parameters of model 2 reconstruction results of different algorithms

Table 4. Geometric parameters of RCL system

Table 4. Geometric parameters of RCL system

Table 5. Algorithm computational performance comparison

Table 5. Algorithm computational performance comparison

Table 6. Quantitative evaluation parameters with different self-prior weights α

Table 6. Quantitative evaluation parameters with different self-prior weights α

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t0

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t0

Table 6. Quantitative evaluation parameters with different self-prior weights $α$

Table 6. Quantitative evaluation parameters with different self-prior weights $α$

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀

Table 7. Quantitative evaluation metrics under different a priori information introduction timings t₀