• Laser & Optoelectronics Progress
  • Vol. 58, Issue 16, 1615003 (2021)
Xuesong Xu* and Yu Zeng
Author Affiliations
  • School of Electrical and Automation Engineering, East China Jiaotong University, Nanchang, Jiangxi 330013, China
  • show less
    DOI: 10.3788/LOP202158.1615003 Cite this Article Set citation alerts
    Xuesong Xu, Yu Zeng. Visual Simultaneous Localization and Mapping Algorithm Based on Dynamic Target Detection[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1615003 Copy Citation Text show less
    Flow chart of our algorithm
    Fig. 1. Flow chart of our algorithm
    Detection results of the moving target
    Fig. 2. Detection results of the moving target
    Matching results of feature points in static region
    Fig. 3. Matching results of feature points in static region
    Detection results of dynamic targets. (a) Two-frame difference method; (b) four-frame difference method; (c) noise filtering; (d) morphological processing; (e) segmentation result
    Fig. 4. Detection results of dynamic targets. (a) Two-frame difference method; (b) four-frame difference method; (c) noise filtering; (d) morphological processing; (e) segmentation result
    Result of dynamic target detection when the image is blurred. (a) Blurred image; (b) binary image; (c) target detection result; (d) optimized binary image; (e) optimized target detection result
    Fig. 5. Result of dynamic target detection when the image is blurred. (a) Blurred image; (b) binary image; (c) target detection result; (d) optimized binary image; (e) optimized target detection result
    Feature point extraction results of different algorithms. (a) Original image; (b) feature points extracted by ORB-SLAM2 algorithm; (c) dynamic targets detected by our algorithm; (d) feature points extracted by our algorithm
    Fig. 6. Feature point extraction results of different algorithms. (a) Original image; (b) feature points extracted by ORB-SLAM2 algorithm; (c) dynamic targets detected by our algorithm; (d) feature points extracted by our algorithm
    Detection results of different algorithms. (a) Original image; (b) YOLO v3 algorithm; (c) YOLO v3+ORB-SLAM2 algorithm; (d) our algorithm
    Fig. 7. Detection results of different algorithms. (a) Original image; (b) YOLO v3 algorithm; (c) YOLO v3+ORB-SLAM2 algorithm; (d) our algorithm
    Trajectories of different algorithms. (a) ORB-SLAM2; (b) ORB-SLAM2+YOLO v3 algorithm; (c) our algorithm
    Fig. 8. Trajectories of different algorithms. (a) ORB-SLAM2; (b) ORB-SLAM2+YOLO v3 algorithm; (c) our algorithm
    SequenceORB-SLAM2Ref.[6]Ref. [7]Ref. [13]Ref. [19]ORB-SLAM2+YOLO v3Ours
    Walking_rpy0.6260--0.17910.13330.44420.06850.0522
    Walking_xyz0.62950.03710.06010.09320.02470.02470.0157
    Walking_halfsphere0.60010.04090.04890.12520.03030.04550.0363
    Walking_static0.12050.00800.02610.06560.00810.02830.0075
    Sitting_halfsphere0.02250.06200.04320.0470--0.07360.0195
    Sitting_xyz0.02350.01130.03970.0482--0.00910.0087
    Table 1. RMSE of ATE with different algorithms unit: m
    Xuesong Xu, Yu Zeng. Visual Simultaneous Localization and Mapping Algorithm Based on Dynamic Target Detection[J]. Laser & Optoelectronics Progress, 2021, 58(16): 1615003
    Download Citation