Indoor Human Motion Recognition Method Based on Kernel-Distance Doppler Velocity Estimation and Lightweight Network

Through-the-wall radar (TWR) achieves human motion recognition by extracting Doppler and micro-Doppler features. In order to solve the problems of the excessive number of parameters in traditional recognition methods and the long inference time of human motion recognition models, this paper proposes a multi-scale indoor human motion recognition method using kernel distance-based Doppler velocity estimation and a lightweight network. The method first utilizes adaptive background subtraction to achieve clutter suppression. To get the micro-Doppler features of human motion, the algorithm utilizes a kernel distance-based Doppler velocity estimation method. In addition, an efficient lightweight target detection neural network is employed to obtain Doppler features of human motion. All the extracted Doppler and micro-Doppler features are combined together for the final decision. The effectiveness of the algorithm is verified through experiments, and the results show that the proposed method can significantly reduce the number of parameters while retaining high recognition accuracy, which shows great potential in real-time recoanition.