Color 3D Reconstruction for Used Machinery Parts using Improved SGM

Zelin Zhang; Xing Cao; Lei Wang; Xuhui Xia

doi:10.3788/LOP231667

Laser & Optoelectronics Progress, Volume. 61, Issue 12, 1211003(2024)

Color 3D Reconstruction for Used Machinery Parts using Improved SGM

Zelin Zhang^1,2, Xing Cao¹, Lei Wang^1,2、*, and Xuhui Xia^1,2

¹Key Laboratory of Metallurgical Equipment and Its Control, Ministry of Education, Wuhan University of Science and Technology, Wuhan 430081, Hubei, China

²Hubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan 430081, Hubei, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(18)

Fig. 1. Color 3D reconstruction process

Download full size

Fig. 2. Polaroid correction diagram

Download full size

Fig. 3. Flowchart of the proposed algorithm

Download full size

Fig. 4. Subpixel coordinate

Download full size

Fig. 5. Cross region

Download full size

Fig. 6. Distance weighting

Download full size

Fig. 7. Principle of triangulation measurement

Download full size

Fig. 8. Binocular vision detection platform

Download full size

Fig. 9. Four kinds of parts with different textures and structures

Download full size

Fig. 10. Image enhancement results before and after enhancement. (a) Part binocular images; (b) binocular images of the parts after image enhancement

Download full size

Fig. 11. Correction results of part image. (a) Before correction;(b) after correction

Download full size

Fig. 12. Parallax map results of different algorithms. (a) SGBM algorithm; (b) TAD-Census algorithm; (c) TFM algorithm; (d) proposed algorithm

Download full size

Fig. 13. Color 3D models generated by different algorithms. (a) Original images;(b) SGBM algorithm;(c) TAD-Census algorithm;(d) TFM algorithm; (e) proposed algorithm

Download full size

Fig. 14. Size reference object

Download full size

Table 1. Calibration results of binocular camera

View table

Table 1. Calibration results of binocular camera

Camera parameter	Left camera	Right camera
Internal parameter matrix	$[\begin{matrix} 1734.454 & 0 & 643.313 \\ 0 & 1733.655 & 513.149 \\ 0 & 0 & 1 \end{matrix}]$	$[\begin{matrix} 1741.547 & 0 & 654.443 \\ 0 & 1740.947 & 540.734 \\ 0 & 0 & 1 \end{matrix}]$
R	$[\begin{matrix} 0.9982 & - 0.0054 & - 0.0593 \\ 0.0053 & 0.9999 & - 0.0007 \\ 0.0593 & - 0.0005 & 0.9982 \end{matrix}]$
T	$[\begin{matrix} - 96.967 & 0.721 & - 0.939 \end{matrix}]$
Distortion coefficient	$C_{l} = [\begin{matrix} \begin{matrix} - 0.101 & 0.194 \end{matrix} & - 0.002 & - 0.001 & 0.915 \end{matrix}]$ $C_{r} = [\begin{matrix} \begin{matrix} - 0.093 & 0.224 \end{matrix} & - 0.001 & 0.001 & - 0.621 \end{matrix}]$

Table 2. Algorithm parameters in this paper
View table
Table 2. Algorithm parameters in this paper
Parameter $λ_{b t}$ $λ_{s}$ $λ_{g}$ $α$ $β$ $γ$ $τ_{1}$ $τ_{2}$ $θ$ $L_{1}$ $L_{2}$
Value 30 30 100 0.88 0.1 0.02 15 20 40 12 16

Table 3. Comparison results of SSIM and MSE for each algorithm

View table

Table 3. Comparison results of SSIM and MSE for each algorithm

Category	SSIM				MSE
Category	SGBM	TAD-Census	TFM	Proposed algorithm	SGBM	TAD-Census	TFM	Proposed algorithm
Part 1	0.226	0.413	0.398	0.493	0.412	0.316	0.287	0.222
Part 2	0.232	0.474	0.584	0.617	0.498	0.309	0.261	0.163
Part 3	0.393	0.492	0.487	0.591	0.397	0.324	0.258	0.184
Part 4	0.298	0.613	0.622	0.673	0.486	0.228	0.204	0.133

Table 4. Comparison of size measurement results of each algorithm

View table

Table 4. Comparison of size measurement results of each algorithm

Algorithm

Actual size /mm

Measured size /mm

Error /mm

Relative error /%

SGBM

77.46

124.26

17.13

2.46

2.26

2.13

3.28

1.85

14.2

TAD-Census

75.97

123.16

16.21

0.97

1.16

1.21

1.29

0.95

8.06

TFM

76.02

123.21

15.9

1.02

1.21

0.90

1.36

0.9

Proposed algorithm

75.61

122.99

15.21

0.46

0.99

0.21

0.81

0.78

1.46

Tools

Get Citation

Copy Citation Text

Zelin Zhang, Xing Cao, Lei Wang, Xuhui Xia. Color 3D Reconstruction for Used Machinery Parts using Improved SGM[J]. Laser & Optoelectronics Progress, 2024, 61(12): 1211003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites