Laser & Optoelectronics Progress, Volume. 59, Issue 16, 1633001(2022)

Stereo Matching Algorithm Based on Multi-Attention Mechanism

Qibo Chen1...2, Baozhen Ge1,2,*, Yunpeng Li1,2, and Jianing Quan12 |Show fewer author(s)
Author Affiliations
  • 1School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China
  • 2Key Laboratory of Opto-Electronics Information Technology, Ministry of Education, Tianjin University, Tianjin 300072, China
  • show less
    Figures & Tables(9)
    Architecture overview of PSMnet
    Architecture overview of proposed algorithm
    Location-based channel attention module
    Multi-cross-attention module
    Attention hourglass module
    Comparison of ReLU function and Mish function
    Parallax evaluation results on KITTI2015 test set. (a1) (a3) Left images; (a2) (a4) partial enlargement of left images; (b1) (b3) disparity maps predicted by PSMnet; (b2) (b4) partial enlargement of disparity maps; (c1) (c3) the corresponding error maps; (c2) (c4) local enlargement of the corresponding error maps; (d1) (d3) disparity maps predicted by MAnet; (d2) (d4) partial enlargement of disparity maps; (e1) (e3) the corresponding error maps; (e2) (e4) partial enlargement of the corresponding error maps
    • Table 1. Performance evaluation of different algorithms on the KITTI2015 test set

      View table

      Table 1. Performance evaluation of different algorithms on the KITTI2015 test set

      AlgorithmD1-bg /%D1-fg /%D1-all /%
      NocAllNocAllNocAll
      GC-NET42.022.215.586.162.612.87
      SegStereo221.761.883.704.072.082.25
      PSMnet51.711.864.314.622.142.32
      Bi3D231.791.953.113.482.012.21
      CTFnet61.681.843.694.032.012.20
      DeepPruner-Best241.711.873.183.561.952.15
      GWCnet251.611.743.493.931.922.11
      SSPCVNet261.611.753.403.891.912.11
      MAnet1.511.653.884.131.902.06
    • Table 2. Performance evaluation of different algorithms on the KITTI2012 test set

      View table

      Table 2. Performance evaluation of different algorithms on the KITTI2012 test set

      Algorithm2-pixels /%3-pixels /%4-pixels /%EPE /pixel
      NocAllNocAllNocAllNocAll
      GC-NET42.713.461.772.301.361.770.60.7
      SegStereo222.663.191.682.031.251.520.50.6
      PSMnet52.443.011.491.891.121.420.50.6
      SSPCVNet262.473.091.471.901.081.410.50.6
      MAnet2.342.971.441.871.091.420.50.6
    Tools

    Get Citation

    Copy Citation Text

    Qibo Chen, Baozhen Ge, Yunpeng Li, Jianing Quan. Stereo Matching Algorithm Based on Multi-Attention Mechanism[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1633001

    Download Citation

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

    Category: Vision, Color, and Visual Optics

    Received: Jun. 9, 2021

    Accepted: Jun. 27, 2021

    Published Online: Jul. 22, 2022

    The Author Email: Baozhen Ge (gebz@tju.edu.cn)

    DOI:10.3788/LOP202259.1633001

    Topics