• Laser & Optoelectronics Progress
  • Vol. 55, Issue 2, 021501 (2018)
Meng Yuan*, Aihua Li, Yong Zheng, Zhigao Cui, and Zhengqiang Bao
Author Affiliations
  • Institute of War Support, Rocket Force University of Engineering, Xi'an, Shaanxi 710025, China
  • show less
    DOI: 10.3788/LOP55.021501 Cite this Article Set citation alerts
    Meng Yuan, Aihua Li, Yong Zheng, Zhigao Cui, Zhengqiang Bao. Point-Line Feature Fusion in Monocular Visual Odometry[J]. Laser & Optoelectronics Progress, 2018, 55(2): 021501 Copy Citation Text show less
    Flowchart of monocular visual odometry
    Fig. 1. Flowchart of monocular visual odometry
    Feature extraction of points (green) and lines (red) in Tum dataset
    Fig. 2. Feature extraction of points (green) and lines (red) in Tum dataset
    Schematic of error analysis for different initialization models
    Fig. 3. Schematic of error analysis for different initialization models
    Schematic of depth filter
    Fig. 4. Schematic of depth filter
    Real-time operation effects in part of the Euroc datasets. (a) MH_01 dataset; (b) MH_02 dataset
    Fig. 5. Real-time operation effects in part of the Euroc datasets. (a) MH_01 dataset; (b) MH_02 dataset
    Location effect in partial Euroc datasets. (a) MH_01 dataset; (b) MH_02 dataset
    Fig. 6. Location effect in partial Euroc datasets. (a) MH_01 dataset; (b) MH_02 dataset
    Real-time operation effects in part of the Tum datasets. (a) fr2_desk dataset; (b) fr2_xyz dataset
    Fig. 7. Real-time operation effects in part of the Tum datasets. (a) fr2_desk dataset; (b) fr2_xyz dataset
    Location effects in partial Euroc datasets. (a) fr2_desk dataset; (b) fr2_xyz dataset
    Fig. 8. Location effects in partial Euroc datasets. (a) fr2_desk dataset; (b) fr2_xyz dataset
    Physical map of robot platform
    Fig. 9. Physical map of robot platform
    Running effects of actual robot. (a) Indoor scene; (b) corridor scene
    Fig. 10. Running effects of actual robot. (a) Indoor scene; (b) corridor scene
    Experiment 1Timeconsuming /msError /pixel
    Line featurealignment1.7370.0059
    Point featureallignment1.8151.2912
    Table 1. Comparison of correction performance for point-line features
    Experiment 2Timeconsuming /msConvergencenumberAverageerror ofdepth value /cm
    Depthestimation ofline feature215.335618.25
    Depthestimation ofpoint feature277.9828615.89
    Depthestimation ofpoint-linefeature347.9734216.27
    Table 2. Comparison of depth estimation performance for point-line features
    Experiment 3RMSE /m
    fc_monoSVOLSD-SLAM(no loop)
    MH_010.160.170.18
    MH_020.210.270.56
    MH_031.700.432.69
    MH_042.711.362.13
    Vicon1_010.820.201.24
    Table 3. Error comparison experiment in Euroc dataset
    Experiment 4RMSE /cm
    fr2_deskfr2_xyz
    fc_mono8.71.4
    SVO (with edgelets)9.71.1
    Semi-dense VO13.53.8
    Feature-based RGB-D SLAM9.52.6
    LSD-SLAM4.51.5
    Table 4. Error comparison experiment in Tum dataset
    Meng Yuan, Aihua Li, Yong Zheng, Zhigao Cui, Zhengqiang Bao. Point-Line Feature Fusion in Monocular Visual Odometry[J]. Laser & Optoelectronics Progress, 2018, 55(2): 021501
    Download Citation