A Multi-feature Fusion Temporal Neural Network for Multi-hand Gesture Recognition using Millimeter-wave Radar Sensor

This paper proposes a multi-feature fusion method for multi-hand gesture recognition method using frequency modulated continuous wave(FMCW) radar sensor. First, the FMCW radar is applied to acquire the intermediate frequency (IF) signals of multi-hand gesture, and the two dimensions fast Fourier transform (2D-FFT) and Multiple Signal classification (MUSIC) algorithm are, respectively, applied to construct the range-Doppler mapping (RDM) and angle-time mapping (ATM) of multi-hand gestures. Then, the interferences in RDM and ATM caused by radar system and the external environment are suppressed. Finally, the multi-feature fusion temporal neural network is proposed to extract the temporal and spatial change information of multi-hand gesture. The experiments are carried out to validate the effectiveness of the proposed method, and the results show that the average recognition accuracy of the proposed method is 97.17%.