• Laser & Optoelectronics Progress
  • Vol. 58, Issue 4, 0410011 (2021)
Dahui Qin, Dong Cheng*, Mingzhu Su, Yunfei Duan, and Yongbo Shao
Author Affiliations
  • School of Civil Engineering and Geomatics, Southwest Petroleum University, Chengdu, Sichuan 610500, China
  • show less
    DOI: 10.3788/LOP202158.0410011 Cite this Article Set citation alerts
    Dahui Qin, Dong Cheng, Mingzhu Su, Yunfei Duan, Yongbo Shao. Photogrammetric Coded Point Localization Based on Target Detection Network[J]. Laser & Optoelectronics Progress, 2021, 58(4): 0410011 Copy Citation Text show less
    Coded markers. (a) Structure of coding; (b) scale of cross coding
    Fig. 1. Coded markers. (a) Structure of coding; (b) scale of cross coding
    Detection flow of YOLO v3
    Fig. 2. Detection flow of YOLO v3
    Bottleneck layer
    Fig. 3. Bottleneck layer
    Transition layer
    Fig. 4. Transition layer
    Dense residual block
    Fig. 5. Dense residual block
    Backbone network of improved YOLO v3
    Fig. 6. Backbone network of improved YOLO v3
    False mark points
    Fig. 7. False mark points
    Center positioning of coded points. (a) Target segmentation; (b) pretreatment; (c) centroid extraction; (d) center positioning
    Fig. 8. Center positioning of coded points. (a) Target segmentation; (b) pretreatment; (c) centroid extraction; (d) center positioning
    Precision-Recall curves of model
    Fig. 9. Precision-Recall curves of model
    Comparison of detection effect of two experiments for the same photo. (a) Experiment 1; (b) experiment 1
    Fig. 10. Comparison of detection effect of two experiments for the same photo. (a) Experiment 1; (b) experiment 1
    Test results in different environments
    Fig. 11. Test results in different environments
    Nearest distance comparison
    Fig. 12. Nearest distance comparison
    Influence of salt-and-pepper noise on recognition rate
    Fig. 13. Influence of salt-and-pepper noise on recognition rate
    Influence of Gaussian noise on recognition rate
    Fig. 14. Influence of Gaussian noise on recognition rate
    Decoding results. (a) Poor condition; (b) good condition
    Fig. 15. Decoding results. (a) Poor condition; (b) good condition
    AlgorithmXTPXFPXFNPRAP /%
    YOLO v397817260.980.9687.86
    M-YOLO10089170.990.9894.91
    Table 1. Detection results of two models
    Dahui Qin, Dong Cheng, Mingzhu Su, Yunfei Duan, Yongbo Shao. Photogrammetric Coded Point Localization Based on Target Detection Network[J]. Laser & Optoelectronics Progress, 2021, 58(4): 0410011
    Download Citation