Incomplete Laser 3D Point Cloud Classification and Completion Approach Based on Object Symmetry under Occlusion Conditions

Yong Tong; Fangyong Xu; Ning Yang; Hui Chen

doi:10.3788/LOP222968

Laser & Optoelectronics Progress, Volume. 60, Issue 20, 2010004(2023)

Incomplete Laser 3D Point Cloud Classification and Completion Approach Based on Object Symmetry under Occlusion Conditions

Yong Tong¹, Fangyong Xu², Ning Yang¹, and Hui Chen^1、*

Author Affiliations

¹College of Automation Engineering, Shanghai University of Electric Power, Shanghai 200090, China

²Changfeng County Power Supply Company, State Grid Anhui Electric Power Co., Ltd., Hefei 231100, Anhui , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(17)

Fig. 1. Approach architecture and system construction. (a) Proposed method framework; (b) schematic of system construction

Download full size

Fig. 2. Partial complete point cloud after multi-station collection and fusion

Download full size

Fig. 3. Schematic of establishing a symmetric plane of incomplete point cloud

Download full size

Fig. 4. Redundant point removal algorithm

Download full size

Fig. 5. Completion results of half residual defect point cloud. (a) Rectangular pyramid; (b) hexagonal pyramid

Download full size

Fig. 6. Completion results of incomplete point cloud less than half (single acquisition). (a) Iron tower;（b）pyramid；（c）dodecahedron；（d）hexagonal prism

Download full size

Fig. 7. Completion results of incomplete point cloud less than half (self defect). (a) House; (b) missile car

Download full size

Fig. 8. Completion results of incomplete point cloud more than half. (a) Cross; (b) sofa

Download full size

Fig. 9. Completion results of incomplete point cloud more than half (cross)

Download full size

Fig. 10. Fusion results of extremely defective point cloud

Download full size

Fig. 11. Completion results of extremely defective point cloud

Download full size

Table 1. System equipment and parameter setting
View table
Table 1. System equipment and parameter setting
Name Parameter setting
Point cloud acquisition device Zivid One⁺ Medium
Optimal acquisition range /m 1‒3
Distance of acquisition d /m 2
Angle of acquisition $α$ /（°） 45

Table 2. Classification of incomplete point cloud

View table

Table 2. Classification of incomplete point cloud

Category	Sample graph	Feature and definition
Half residual defect point cloud		or ，and at least
Incomplete point clouds less than half		，，
Incomplete point clouds more than half		∀Di>0 or ∀Di<0，min{n,m}=0，max{n,m}>0，100σ≥min{Di}>σ or less than N10000Di≤σ2
Extremely defective point cloud		or ，，，and

Table 3. Evaluation index value of half residual defect point cloud

View table

Table 3. Evaluation index value of half residual defect point cloud

Evaluation index	Hexagonal pyramid	Rectangular pyramid
RMS /cm	0.1028	0.1702
Surface area of half residual defect point cloud /cm²	113.17	188.51
Surface area of the full model point cloud /cm²	200.81	327.90
Surface area of the fused point cloud /cm²	199.40	327.09
Surface area approximation of the fused point cloud to the full model point cloud /%	99.3	99.8

Table 4. Evaluation index value of incomplete point cloud less than half

View table

Table 4. Evaluation index value of incomplete point cloud less than half

Evaluation index	Iron tower	Pyramid	Dodecahedron	Hexagonal prisms	House	Missile car
RMS /cm	0.1379	0.0806	0.0867	0.0945	0.1746	0.1288
Surface area of incomplete point cloud less than half /cm²	67.56	71.02	124.04	209.92	122.45	56.92
Surface area of the full model point cloud /cm²	117.51	138.74	219.13	275.73	134.78	75.06
Surface area of the fused point cloud /cm²	117.29	136.79	217.51	270.67	134.31	73.72
Surface area approximation of the fused point cloud to the full model point cloud /%	99.8	98.6	99.3	98.2	99.7	98.2

Table 5. Evaluation index value of incomplete point cloud more than half

View table

Table 5. Evaluation index value of incomplete point cloud more than half

Evaluation index	Cross	Sofa
RMS /cm	0.1484	0.1382
Surface area of incomplete point cloud more than half /cm²	148.77	21225.35
Surface area of the full model point cloud /cm²	356.26	50048.46
Surface area of the fused point cloud /cm²	362.06	50318.77
Surface area approximation of the fused point cloud to the full model point cloud /%	98.4	99.5

Table 6. Evaluation index value of extremely defective point cloud

View table

Table 6. Evaluation index value of extremely defective point cloud

Evaluation index	Faucet	Armchair
RMS /cm	0.0549	0.1210
Surface area of extremely defective point cloud /cm²	630.03	10330.24
Surface area of the full model point cloud /cm²	2051.74	31776.47
Surface area of the fused point cloud /cm²	2082.19	31079.61
Surface area approximation of the fused point cloud to the full model point cloud /%	98.5	97.8

Tools

Get Citation

Copy Citation Text

Yong Tong, Fangyong Xu, Ning Yang, Hui Chen. Incomplete Laser 3D Point Cloud Classification and Completion Approach Based on Object Symmetry under Occlusion Conditions[J]. Laser & Optoelectronics Progress, 2023, 60(20): 2010004

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites