Laser & Optoelectronics Progress, Volume. 56, Issue 3, 031010(2019)

Obstacle Recognition in Vegetation Environment Based on Markov Random Field

Ziyang Cheng, Guoquan Ren*, and Yin Zhang
Author Affiliations
  • Department of Vehicle and Electrical Engineering, Army Engineering University, Shijiazhuang, Hebei 050003, China
  • show less
    Figures & Tables(15)
    Connections between adjacent points. (a) Simplest connection layout; (b) extension of distance between connected points
    Principle diagram of threshold t2
    Expansion of connections. (a) Point cloud; (b) untypical connections; (c) connections after expansion
    Parameters of neighborhood characteristic. (a) θV; (b) θL; (c) θP
    Parameter distribution characteristics. (a) θV; (b) θL; (c) θP; (d) θF
    Flow chart of algorithm
    Ground unmanned experimental platform
    Experimental scene for parameter selection
    Results of different values of δ. (a) δ=0.4; (b) δ=0.6; (c) δ=0.8; (d) δ=1.0
    Experimental scenes. (a) Scene 1; (b) scene 2
    Detection results in artificial scene. (a) Manual marking; (b) proposed algorithm; (c) model matching algorithm; (d) covariance algorithm
    Detection results in field scene. (a) Manual marking; (b) proposed algorithm; (c) model matching algorithm; (d) covariance algorithm
    Comparison of running time of different algorithms. (a) Artificial scene; (b) field scene
    • Table 1. Parameter selection

      View table

      Table 1. Parameter selection

      δ0.40.60.81.0
      Number ofmisclassified points156852474
      Correspondinggraphic numberFig. (a)Fig. (b)Fig. (c)Fig. (d)
    • Table 2. Result comparison of detection algorithms in different scenes

      View table

      Table 2. Result comparison of detection algorithms in different scenes

      SceneAlgorithmTPR /%FPR /%
      Artificial sceneModel matching algorithmCovariance algorithmProposed algorithm84.6824.63
      82.1919.81
      93.524.26
      Field sceneModel matching algorithmCovariance algorithmProposed algorithm78.7329.21
      62.9234.46
      89.948.13
    Tools

    Get Citation

    Copy Citation Text

    Ziyang Cheng, Guoquan Ren, Yin Zhang. Obstacle Recognition in Vegetation Environment Based on Markov Random Field[J]. Laser & Optoelectronics Progress, 2019, 56(3): 031010

    Download Citation

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

    Category: Image Processing

    Received: Jul. 30, 2018

    Accepted: Aug. 31, 2018

    Published Online: Jul. 31, 2019

    The Author Email: Guoquan Ren (rrrgggqqq@163.com)

    DOI:10.3788/LOP56.031010

    Topics