Target Localization and Tracking Method Based on Camera and LiDAR Fusion

Pu Zhang; Jinqing Liu; Jinchao Xiao; Junfeng Xiong; Tianwei Feng; Zhongze Wang

doi:10.3788/LOP231537

Laser & Optoelectronics Progress, Volume. 61, Issue 8, 0828004(2024)

Target Localization and Tracking Method Based on Camera and LiDAR Fusion

Pu Zhang^1,2,3,4, Jinqing Liu^1,2,3、*, Jinchao Xiao⁴, Junfeng Xiong⁴, Tianwei Feng⁴, and Zhongze Wang⁴

¹Key Laboratory of Optoelectronic Science and Technology for Medicine of Ministry of Education, Fujian Normal University, Fuzhou 350007, Fujian, China

²Fujian Provincial Key Laboratory of Photonic Technology, Fujian Normal University, Fuzhou 350007, Fujian, China

³Fujian Provincial Engineering Technology Research Center of Photoelectric Sensing Application, Fujian Normal University, Fuzhou 350007, Fujian, China

⁴Guangzhou Institute of Industrial Intelligence, Guangzhou 511458, Guangdong, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. The process of camera and LiDAR fusion algorithm

Download full size

Fig. 2. Camera and LiDAR coordinate system transformation

Download full size

Fig. 3. Point cloud screening

Download full size

Fig. 4. LiDAR point cloud projection

Download full size

Fig. 5. Target point cloud selection. (a) Current point cloud cluster selection; (b) current point cloud cluster projection; (c) selection of remaining point cloud clusters; (d) current point cloud cluster selection; (e) current point cloud cluster projection; (f) selection of remaining point cloud clusters

Download full size

Fig. 6. Process of obstacle point cloud filtering algorithm based on area comparison

Download full size

Fig. 7. The result of kitti datatest. (a) Image detection; (b) before clustering; (c) after clustering; (d) 3D bounding box

Download full size

Fig. 8. Target positioning comparison

Download full size

Fig. 9. Occluded target tracking rendering

Download full size

Fig. 10. Target tracking results. (a) X-trace; (b) Y-trace

Download full size

Table 1. Point cloud components within the 2D box
View table
Table 1. Point cloud components within the 2D box
Case 1 Case 2 Case 3 Case 4
Target point cloud
Target point cloud；
obstacle point cloud
Target point cloud；background point cloud
Target point cloud；
obstacle point cloud；
background point cloud

Table 2. Threshold comparison
View table
Table 2. Threshold comparison
$σ$ 1/2 7/12 2/3 3/4
Accuracy /% 81.1250 85.4167 88.5417 74.4791

Table 3. Algorithm comparison

View table

Table 3. Algorithm comparison

Algorithm	Easy /%	Moderate /%	Hard /%	Accuracy /%	Time /s
Algorithm of reference［7］	82.6667	86.9565	58.3333	78.125	0.1218
Algorithm of reference［11］	84	95.6521	52.0833	80.2083	0.1299
Proposed algorithm	92	97.1014	70.8333	88.5417	0.0314

Table 4. Comparison of average errors
View table
Table 4. Comparison of average errors
Algorithm e_x /pixel e_y /pixel o_r /% Time /s
PF 14.7 5.87 23.4 0.1175
DeepSORT 9.45 2.17 87.06 0.0817
Proposed algorithm 4.49 1.80 87.42 0.0841

Tools

Get Citation

Copy Citation Text

Pu Zhang, Jinqing Liu, Jinchao Xiao, Junfeng Xiong, Tianwei Feng, Zhongze Wang. Target Localization and Tracking Method Based on Camera and LiDAR Fusion[J]. Laser & Optoelectronics Progress, 2024, 61(8): 0828004

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites