Infrared and Laser Engineering, Volume. 52, Issue 3, 20220618(2023)
Pose estimation of flying target based on bi-modal information fusion
Figures & Tables(29)
Table 1. Parameter range of RS-LD1605 M radar
View tableView in ArticleTable 1. Parameter range of RS-LD1605 M radar
Name Value Model RS-LD1605 M Horizontal field of view/(°) 56 Vertical field of view/(°) 31 Color image resolution 2448×1378 Depth image resolution 640×360 Horizontal angle resolution/(°) 0.088 Vertical angle resolution/(°) 0.088 Detection distance/m 50 Detection accuracy/cm 5 Frame rate/Hz 12
Table 2. Comparison of algorithm accuracy
View tableView in ArticleTable 2. Comparison of algorithm accuracy
Group ΔPx/mm ΔPy/mm ΔPz/mm Δβ/(°) Δ γ /(°)Time/ms Proposed algorithm 1.06 4.59 2.07 0.63 1.01 132 ICP algorithm 2.93 8.37 4.21 0.94 1.62 261 PnP algorithm 1.65 5.45 4.11 0.76 1.33 37 Algorithm in Ref.[ 10 ]2.51 6.02 3.88 0.82 1.41 327
Table 3. Comparison of target extraction accuracy of image
View tableView in ArticleTable 3. Comparison of target extraction accuracy of image
Scene Group TP FP TN FN P R PWC Scene A Our method 1 520 40 336 968 4 280 0.962 0.844 0.009 4% Background subtraction 1 627 6 015 336 372 9 173 0.213 0.904 0.183% Vibe algorithm 1 642 178 336 956 6 158 0.902 0.912 0.009 9% Algorithm in Ref. [ 5 ]1 732 117 336 962 7 68 0.936 0.962 0.005 5% Scene B Our method 1 607 31 336 931 3 393 0.978 0.803 0.013% Background subtraction 1 627 7 302 336 204 2 373 0.182 0.813 0.228% Vibe algorithm 1 766 725 336 861 9 234 0.709 0.883 0.028% Algorithm in Ref. [ 5 ]1 939 1 007 336 833 7 61 0.658 0.969 0.032%
Table 4. Time consuming for image target extraction
View tableView in ArticleTable 4. Time consuming for image target extraction
Group A/ms B/ms Proposed algorithm 301 289 Background subtraction 153 147 Vibe algorithm 271 253 Algorithm in Ref.[ 5 ]206 198
Table 5. Calculation results of target pose
View tableView in ArticleTable 5. Calculation results of target pose
Scene number ΔPx/mm ΔPy/mm ΔPz/mm Δβ/(°) Δ γ /(°)Scene A 2.7015 −0.4149 10.2778 43.6813 66.7326 Scene B 1.3368 0.1723 10.5036 82.8329 4.6933
Table 6. PnP pose calculation
View tableView in ArticleTable 6. PnP pose calculation
Scene number ΔPx/mm ΔPy/mm ΔPz/mm Δβ/(°) Δ γ /(°)Scene A 2.7073 −0.4213 10.2818 44.3518 65.3863 Scene B 1.3328 0.1685 10.5072 81.7567 5.3052 Scene A 2.7062 −0.4137 10.2747 44.2721 65.8147 Scene B 1.3317 0.1705 10.5093 81.2602 5.3872
Table 7. Position and attitude error comparison
View tableView in ArticleTable 7. Position and attitude error comparison
Scene number ΔPx/mm ΔPy/mm ΔPz/mm Δβ/(°) Δ γ /(°)Scene A 5.25 2.6 4.5 0.63 −1.13 Scene B −4.55 −2.8 4.65 −1.32 0.65
Table 8. Point cloud collection rate under different surface colors
View tableView in ArticleTable 8. Point cloud collection rate under different surface colors
Colour White Blue Red Acquisition rate 1 0.97 0.94
Table 9. Number of point clouds at different object distances and surface colors
View tableView in ArticleTable 9. Number of point clouds at different object distances and surface colors
Group ΔPx/mm ΔPy/mm ΔPz/mm Δβ/(°) Δ γ /(°)Group A 0.97 4.26 1.84 0.58 0.97 Group B 0.061 0.062 0.039 0.014 0.082 Our algorithm 1.06 4.59 2.07 0.63 1.01
Table 10. Position error
View tableView in ArticleTable 10. Position error
Group ΔPx/mm ΔPy/mm ΔPz/mm Group A 0.97 4.26 1.84 Group B 0.061 0.062 0.039 Our method 1.06 4.59 2.07
Table 11. Angle error
View tableView in ArticleTable 11. Angle error
Group Δβ/(°) Δγ/(°) Group A 0.58 0.97 Group B 0.014 0.082 Our method 0.63 1.01
Tools
Get Citation
Copy Citation Text
Ronghua Li, Meng Wang, Wei Zhou, Jiaru Fu. Pose estimation of flying target based on bi-modal information fusion[J]. Infrared and Laser Engineering, 2023, 52(3): 20220618
Paper Information
Category: Photoelectric measurement
Received: Aug. 10, 2022
Accepted: --
Published Online: Apr. 12, 2023
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20