• Laser & Optoelectronics Progress
  • Vol. 58, Issue 22, 2207002 (2021)
Yao Li*, Xin Wang, Wentao He, and Baodai Shi
Author Affiliations
  • Tracking Guidance Teaching and Research Section, Air Defense and Missile Defense College, Air Force Engineering University, Xi’an, Shaanxi 710051, China
  • show less
    DOI: 10.3788/LOP202158.2207002 Cite this Article Set citation alerts
    Yao Li, Xin Wang, Wentao He, Baodai Shi. Hand Gesture Recognition Using Ultra-Wideband Radar with Random Forest[J]. Laser & Optoelectronics Progress, 2021, 58(22): 2207002 Copy Citation Text show less
    Flow chart of echo signal processing
    Fig. 1. Flow chart of echo signal processing
    Comparison of time domain and frequency domain among different gestures. (a) Gesture 1; (b) gesture 2; (c) gesture 3; (d) gesture 4
    Fig. 2. Comparison of time domain and frequency domain among different gestures. (a) Gesture 1; (b) gesture 2; (c) gesture 3; (d) gesture 4
    Flowchart of system working
    Fig. 3. Flowchart of system working
    Feature map clustering analysis
    Fig. 4. Feature map clustering analysis
    Clustering results for different K values. (a) Original image; (b) K=2; (c) K =4; (d) K =8
    Fig. 5. Clustering results for different K values. (a) Original image; (b) K=2; (c) K =4; (d) K =8
    Visualization of clustering results
    Fig. 6. Visualization of clustering results
    Random forest visualization
    Fig. 7. Random forest visualization
    Contribution rate of some features
    Fig. 8. Contribution rate of some features
    SNR of maps with different channel numbers
    Fig. 9. SNR of maps with different channel numbers
    Number of accumulated echo pulses4896128256
    Average accuracy of classification /%67.2381.4698.6799.97
    Table 1. Comparison of recognition accuracy for different number of accumulated echo pulses
    Type of feature mapAverage accuracy of classification /%
    RGB map98.67
    R channel map77.98
    G channel map81.25
    B channel map78.25
    Gray-scale map83.54
    Table 2. Comparison of accuracy for different feature maps
    Type of inputCART numberMaximum number of featuresAverage accuracy of classification /%
    RDM50679.26
    1285.32
    2086.17
    100687.34
    1298.67
    2099.12
    200686.54
    1297.12
    2098.31
    Table 3. Results of classification for different parameter settings of random forest
    ParameterSetting
    Number of transmitting antennas1
    Number of receiving antennas1
    Number of integrated radar pulses128
    RDM frame duration /ms14
    Number of sampling points8000
    Type of inputRDM
    Size of input512×512×3
    CART number100
    Maximum number of features12
    Number of gesture classes6
    Table 4. Setting of parameters in gesture recognition algorithm
    AlgorithmTraining time /hRecognition speed /(frame·s-1)Accuracy /%
    ShuffleNet V24.464190.64
    Mobilenet V24.524291.53
    VGG 164.214686.63
    Algorithm proposed in this article2.614198.93
    Table 5. Comparison of proposed algorithm with other algorithms
    Yao Li, Xin Wang, Wentao He, Baodai Shi. Hand Gesture Recognition Using Ultra-Wideband Radar with Random Forest[J]. Laser & Optoelectronics Progress, 2021, 58(22): 2207002
    Download Citation