Efficient Filtering and Smoothing Algorithm For Train Key Components Based on Scattered Point Clouds

Ni Zeng; Jinlong Li; Xiaorong Gao; Yu Zhang; Lin Luo

doi:10.3788/LOP221987

Journals >Laser & Optoelectronics Progress >Volume 60 >Issue 14 >Page 1410011 > Article

Laser & Optoelectronics Progress
Vol. 60, Issue 14, 1410011 (2023)

Efficient Filtering and Smoothing Algorithm For Train Key Components Based on Scattered Point Clouds

Ni Zeng, Jinlong Li^*, Xiaorong Gao, Yu Zhang, and Lin Luo

Author Affiliations

School of Physical Science and Technology, Southwest Jiaotong University, Chengdu 610031, Sichuan, China

show less

DOI: 10.3788/LOP221987 Cite this Article Set citation alerts

Ni Zeng, Jinlong Li, Xiaorong Gao, Yu Zhang, Lin Luo. Efficient Filtering and Smoothing Algorithm For Train Key Components Based on Scattered Point Clouds[J]. Laser & Optoelectronics Progress, 2023, 60(14): 1410011 Copy Citation Text

show less

Fig. 1. Point cloud adaptive weighted guided filtering

Download full size

Fig. 2. Diagram of K-D tree space division

Download full size

Fig. 3. Schematic of KNN algorithm

Download full size

Fig. 4. Comparison of train wheel pair point clouds before and after filtering. (a) Standard point cloud; (b) after applying Gaussian noise; (c) bilateral filtering; (d) guided filtering; (e) adaptive weighted guided filtering

Download full size

Fig. 5. Comparison of train bogie point clouds before and after filtering. (a) Standard point cloud; (b) after applying Gaussian noise; (c) bilateral filtering; (d) guided filtering; (e) adaptive weighted guided filtering

Download full size

Fig. 6. Comparison of train component 1. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Fig. 7. Comparison of train component 2. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Fig. 8. Comparison of train component 3. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Fig. 9. Comparison of train component 4. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Fig. 10. Comparison of train component 5. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Fig. 11. Comparison of train component 6. (a) Original point cloud; (b) bilateral filtering; (c) guided filtering; (d) adaptive weighted guided filtering

Download full size

Train wheel pair	Time /ms	$D_{m a x}$	$D_{m e a n}$	$D_{s t d}$
Gaussian noise		92.69	16.70	10.50
BF	1419.20	59.28	10.75	6.78
GF	560.48	56.45	10.62	6.28
AWGF	715.10	56.79	10.39	6.14

Table 1. Train wheel pair point cloud filtering results

Train bogie	Time /ms	$D_{m a x}$	$D_{m e a n}$	$D_{s t d}$
Gaussian noise		91.10	14.70	10.09
BF	35081.10	84.33	9.70	6.28
GF	15886.60	56.11	9.62	5.90
AWGF	16072.50	55.10	9.29	5.74

Table 2. Train bogie point cloud filtering results

Train key component	Number of points
1	218452
2	332895
3	523009
4	843918
5	1029027
6	1834880

Table 3. Number of point clouds of train components

Train component	1	2	3	4	5	6
BF	10086.2	15451.8	24106.4	32991.5	43347.6	78469.9
GF	4274.4	6579.5	10225.4	16610.5	19983.2	36216.9
AWGF	4376.1	6803.4	10667.7	17129.0	20345.7	37077.5

Table 4. Comparison of filtering time for different number of point cloud components

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information