Neutron Computed Tomography Reconstruction Method Using Sparse Projections Based on Weighted Total Difference Minimization

Qiang Lin; Min Yang; Bin Tang; Bin Liu; Heyong Huo; Jiawei Liu

doi:10.3788/AOS201939.0711003

Acta Optica Sinica, Volume. 39, Issue 7, 0711003(2019)

Neutron Computed Tomography Reconstruction Method Using Sparse Projections Based on Weighted Total Difference Minimization

Qiang Lin¹, Min Yang^1、*, Bin Tang², Bin Liu², Heyong Huo², and Jiawei Liu¹

Author Affiliations

¹ School of Mechanical Engineering and Automation, Beijing University of Aeronautics and Astronautics, Beijing 100191, China

² Institute of Nuclear Physics and Chemistry, China Academy of Engineering Physics, Mianyang, Sichuan 621900, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(15)

Fig. 1. Reconstructed images of Shepp-Logan. (a) Ideal image; (b) ART algorithm; (c) SART algorithm; (d) SIRT algorithm

Download full size

Fig. 2. Reconstructed images of Shepp-Logan using noisy projections. (a) Ideal image; (b) ART algorithm; (c) SART algorithm; (d) SIRT algorithm

Download full size

Fig. 3. Reconstructed images obtained by FBP and SIRT with different sparse angle projections. (a) FBP reconstructed image with 180 projections; (b) FBP reconstructed image with 20 projections; (c) SIRT reconstructed image with 180 projections; (d) SIRT reconstructed image with 20 projections

Download full size

Fig. 4. Difference measures of WTDM method in different directions[17]. (a) Diagram of four neighbors of pixel point; (b) partial gradients of pixel point in different directions

Download full size

Fig. 5. Reconstructed images of Shepp-Logan using 90 projections. (a) FBP reconstruction; (b) SIRT reconstruction; (c) SART-WTDM reconstruction; (d) SIRT-WTDM reconstruction

Download full size

Fig. 6. Reconstructed images of Shepp-Logan using 90 noisy projections. (a) FBP reconstruction; (b) SIRT reconstruction; (c) SART-WTDM reconstruction; (d) SIRT-WTDM reconstruction

Download full size

Fig. 7. Gray curves of reconstructed images of Shepp-Logan with 90 projections. (a) Without noise in projections; (b) with noise in projections

Download full size

Fig. 8. Reconstructed images of code tube sample with 75 projections. (a) FBP reconstruction with complete projections; (b) FBP reconstruction with 75 projections; (c) SIRT reconstruction with 75 projections; (d) SART-WTDM reconstruction with 75 projections; (e) SIRT-WTDM reconstruction with 75 projections

Download full size

Fig. 9. Reconstructed images of code tube sample with 18 projections. (a) FBP reconstruction with complete projections; (b) FBP reconstruction with 18 projections; (c) SIRT reconstruction with 18 projections; (d) SART-WTDM reconstruction with 18 projections; (e) SIRT-WTDM reconstruction with 18 projections

Download full size

Fig. 10. 3D visualization results of code tube sample. (a) Reconstructed region of code tube sample; (b) FBP reconstructed result with complete projections; (c) FBP reconstructed result with 75 projections; (d) SIRT reconstructed result with 75 projections; (e) SART-WTDM reconstructed result with 75 projections; (f) SIRT-WTDM reconstructed result with 75 projections

Download full size

Table 1. MSE and PSNR of reconstructed images of Shepp-Logan

View table

Table 1. MSE and PSNR of reconstructed images of Shepp-Logan

Method	Without noise in the projections		With noise in the projections			Image quality reduction rate
Method	MSE	PSNR /dB	MSE	PSNR /dB		Increasing rate of MSE /%	Reduction rate of PSNR /%
ART	0.0047				71.4516			0.0078	69.1832	65.96	3.17
SART	0.0047				71.4208			0.0077	69.2508	63.83	3.04
SIRT	0.0046				71.5167			0.0049	71.2571	6.52	0.36

Table 2. Simulative CT conditions based on Shepp-Logan
View table
Table 2. Simulative CT conditions based on Shepp-Logan
Scanning mode Reconstructing image size /(pixel×pixel) Range of scanning angle /(°) Angle step /(°)
Parallel beam 256×256 180 2

Table 3. MSE and PSNR of reconstructed images of Shepp-Logan with 90 projections

View table

Table 3. MSE and PSNR of reconstructed images of Shepp-Logan with 90 projections

Method	Without noise in projections		With noise in projections
Method	MSE	PSNR /dB	MSE	PSNR /dB
FBP	0.0100		68.1374		0.0118	67.4080
SIRT	0.0046		71.4593		0.0064	70.0438
SART-WTDM	0.0045		71.5842		0.0062	70.1929
SIRT-WTDM	0.0045		71.5850		0.0062	70.2200

Table 4. CT scanning conditions of code tube sample
View table
Table 4. CT scanning conditions of code tube sample
Neutron beam type Detector type Number of detector pixels Range of scanning angle /(°) Total projection number
Cold neutron Andor iKon-L 936 2048×2048 180 450

Table 5. MSE and PSNR of reconstructed images of code tube sample

View table

Table 5. MSE and PSNR of reconstructed images of code tube sample

Method	776^th slice image			313^rd slice image
Method	MSE	PSNR /dB		MSE	PSNR /dB
FBP	9.8263×10^-4		78.2069			0.0131	66.9626
SIRT	9.4837×10^-4		78.3610			0.0090	68.6035
SART-WTDM	7.6070×10^-4		79.3187			0.0042	71.8748
SIRT-WTDM	4.0869×10^-4		82.0169			0.0037	72.4440

Tools

Get Citation

Copy Citation Text

Qiang Lin, Min Yang, Bin Tang, Bin Liu, Heyong Huo, Jiawei Liu. Neutron Computed Tomography Reconstruction Method Using Sparse Projections Based on Weighted Total Difference Minimization[J]. Acta Optica Sinica, 2019, 39(7): 0711003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites