Laser & Optoelectronics Progress, Volume. 57, Issue 22, 221001(2020)

Tracking of Human Respiratory Motion Based on Kanade-Lucas-Tomasi Algorithm

Xinyu Liu1, Zheng Yan1, Fang Duan1、*, and Zhongying Dai2、*
Author Affiliations
  • 1School of Information Science and Engineering, Huaqiao University, Xiamen, Fujian 361021, China
  • 2Institute of Modern Physics, Chinese Academy of Sciences, Lanzhou, Gansu 730000, China
  • show less
    Figures & Tables(13)
    Schematic diagram of experiment 1
    Schematic diagram of experiment 2
    Flow chart of image pre-processing
    Flow chart of tracking algorithm
    Results of pre-processing. (a) Original image; (b) down-sampled image; (c) image filtered by Lee filter; (d) image filtered by Kuwahara filter; (e) image denoised by Gaussian filter; (f) edge detection result with Canny operator
    Tracking results of experiment 1. (a) Mark four corner points of marker in first frame manually; (b) ROI calculated by algorithm according to marked coordinates of corner points; (c) tracking result of 100th frame with KLT algorithm
    Curves of actual displacement of body surface and pixel difference between ordinate frames of ROI corner points
    Tracking results from three perspectives
    Comparison of phantom motion curves and algorithm tracking curves under three breathing conditions. (a) Three breathing conditions; (b) phantom motion curves; (c) algorithm tracking curves
    Actual displacement of marker and pixel difference between feature points
    Error bar of tracking experimental results
    • Table 1. Table of comparison of actual displacement and pixel difference between frames

      View table

      Table 1. Table of comparison of actual displacement and pixel difference between frames

      Number of image frames12345678910
      Actual displacement /mm0.5900.6370.6840.6840.7300.7300.7770.7770.8240.824
      Pixel number01.05192.32203.99025.74977.25078.13099.444110.554611.7608
      Actual displacement afternormalization00.00230.00470.00470.00700.00700.00940.00940.01180.0118
      Normalized pixel value00.03490.07710.13250.19100.24080.27010.31370.35060.3907
      Normalized actual displacementafter down-sampling0.00770.07670.14000.19670.23340.27920.31680.36490.41900.4641
      Normalized pixel value afterdown-sampling0.00060.03700.08530.14510.20900.25190.28870.33430.37390.4180
      Difference between actualdisplacement and pixel change0.00710.03970.05470.05160.02440.02730.02810.03060.04510.0461
    • Table 2. Comparison of actual displacement and changes in position of central point of ROI

      View table

      Table 2. Comparison of actual displacement and changes in position of central point of ROI

      Number of image frames14080120160200240280320360
      Actual displacement /cm14.1413.6913.5813.9814.0813.6813.5213.8814.0513.75
      Pixel difference /pixel214.73189.56183.00205.52210.38188.63180.66199.61209.55193.61
      Actual displacement A /cm-0.45-0.110.400.10-0.40-0.160.360.17-0.30
      Pixel difference B /pixel-25.18-6.5622.524.86-21.75-7.9718.959.94-15.95
      Normalized A (C)0-0.69-0.86-0.25-0.09-0.71-0.95-0.40-0.14-0.60
      Normalized B (D)0-0.70-0.88-0.25-0.12-0.72-0.94-0.42-0.14-0.58
      Difference between C and D00.010.0200.030.010.010.0200.02
    Tools

    Get Citation

    Copy Citation Text

    Xinyu Liu, Zheng Yan, Fang Duan, Zhongying Dai. Tracking of Human Respiratory Motion Based on Kanade-Lucas-Tomasi Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221001

    Download Citation

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

    Category: Image Processing

    Received: Feb. 5, 2020

    Accepted: Mar. 27, 2020

    Published Online: Nov. 3, 2020

    The Author Email: Fang Duan (nkfetsh@gmail.com), Zhongying Dai (nkfetsh@gmail.com)

    DOI:10.3788/LOP57.221001

    Topics