Laser & Optoelectronics Progress, Volume. 60, Issue 12, 1210020(2023)

3D Recognition Algorithm Based on Curvature Point Pair Features

Menghui Yu1, Xining Cui1, Linqigao Wu1, and Shiqian Wu1,2、*
Author Affiliations
  • 1School of Information Science and Engineering, Wuhan University of Science and Technology, Wuhan 430081, Hubei, China
  • 2Institute of Robotics and Intelligent Systems, Wuhan University of Science and Technology, Wuhan 430081, Hubei, China
  • show less
    Figures & Tables(17)
    Schematic of curvature difference PPF
    Description of global model
    Transformation between model and scene coordinates
    Curvature and curvature histogram of buddha. (a) Curvature of buddha; (b) curvature histogram of buddha
    Visualization of weighted voting process
    Point clouds of models
    Matching effect of four algorithms. (a) Original PPF algorithm; (b) S-ICP algorithm; (c) curvature difference PPF algorithm; (d) curvature difference PPF algorithm with ICP
    Matching effect of four algorithms. (a) Original PPF algorithm; (b) S-ICP algorithm; (c) curvature difference PPF algorithm;(d) curvature difference PPF algorithm with ICP
    Process of scene segmentation. (a) Point cloud of scene; (b) grayscale image of point cloud mapping; (c) watershed segmentation; (d) candidate target
    Matching effect of four algorithms. (a) Original PPF algorithm; (b) S-ICP algorithm; (c) curvature difference PPF algorithm; (d) curvature difference PPF algorithm with ICP
    Intermediate result of point pair matching. (a) A point pair in scene; (b) original PPF algorithm; (c) curvature difference PPF algorithm
    • Table 1. Comparison of matching rates of algorithms

      View table

      Table 1. Comparison of matching rates of algorithms

      Algorithmarmadillobuddhabunnydragonchinese_dragonstatuetteAverage
      PPF86.4090.2591.9392.7885.9084.9188.70
      S-ICP98.6792.0397.7993.9497.6194.6995.79
      Proposed algorithm93.6191.2695.0895.9287.9589.8992.29
      Proposed algorithm+ICP98.9696.0198.1997.8397.8696.8597.62
    • Table 2. Comparison of matching time of algorithms

      View table

      Table 2. Comparison of matching time of algorithms

      Algorithmarmadillobuddhabunnydragonchinese_dragonstatuetteAverage
      PPF0.3010.7750.1360.7120.8280.9340.614
      S-ICP0.4110.5420.5360.7970.4980.5760.560
      Proposed algorithm0.1420.1780.0850.5110.1080.4570.247
      Proposed algorithm+ ICP0.1600.2160.1010.5290.1260.4730.268
    • Table 3. Comparison of matching rates of algorithms

      View table

      Table 3. Comparison of matching rates of algorithms

      AlgorithmbuddhabunnystatuetteAverage
      PPF89.5883.5082.4585.18
      S-ICP93.2493.9894.6493.95
      Proposed algorithm91.8193.3194.1893.10
      Proposed algorithm+ICP96.1796.2396.2296.21
    • Table 4. Comparison of matching time of algorithms

      View table

      Table 4. Comparison of matching time of algorithms

      AlgorithmbuddhabunnystatuetteAverage
      PPF0.7210.0911.0770.630
      S-ICP0.6330.4630.8750.657
      Proposed algorithm0.2540.0440.4700.256
      Proposed algorithm+ICP0.2780.0670.5370.294
    • Table 5. Comparison of matching rates of algorithms

      View table

      Table 5. Comparison of matching rates of algorithms

      Algorithm(1)(2)(3)(4)(5)(6)Average
      PPF80.6257.0978.5480.9184.2186.0577.90
      S-ICP94.3792.2292.8691.3991.9493.4992.71
      Proposed algorithm89.7288.4391.9484.3387.8585.9188.03
      Proposed algorithm+ICP95.9295.6096.3491.6494.7394.3294.76
    • Table 6. Comparison of matching time of algorithms

      View table

      Table 6. Comparison of matching time of algorithms

      AlgorithmPPFS-ICPProposed algorithmProposed algorithm+ICP
      Total time9.9088.3684.8155.311
    Tools

    Get Citation

    Copy Citation Text

    Menghui Yu, Xining Cui, Linqigao Wu, Shiqian Wu. 3D Recognition Algorithm Based on Curvature Point Pair Features[J]. Laser & Optoelectronics Progress, 2023, 60(12): 1210020

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Apr. 6, 2022

    Accepted: Jul. 14, 2022

    Published Online: Jun. 5, 2023

    The Author Email: Shiqian Wu (shiqian.wu@wust.edu.cn)

    DOI:10.3788/LOP221223

    Topics