Gesture recognition with feature fusion using FMCW radar

We propose a feature fusion method for gesture recognition based on Frequency Modulated Continuous Wave (FMCW) radar. First, we estimate the radial information distance, Doppler and tangential information azimuth, elevation of the gesture by signal processing to construct the multi-dimensional feature data set; Then, for feature extraction and accurate classification, we design feature fusion scheme and build multi-dimensional feature convolutional neural network. Experimental results show that our proposed scheme with gesture multi-dimensional feature as input can solve the problem of insufficient feature representation in traditional Range-Doppler (RD) domain gesture recognition methods, and the recognition accuracy is improved by 4%similar to 8% compared with the case without feature fusion.