Feature extraction and classification of human motions with LFMCW radar

The use of micro-Doppler radar signatures for feature extraction and classification of human activities has become a focus recently thanks to the discovering of micro-Doppler effect. Compared with the traditional Doppler radar, LFMCW radar is able to distinguish Multi-objective and detect target range. In this paper, we analyze the difference between running and walking which are based on Boulic model with Short Time Fourier Transform (STFT) in time-frequency domain and give four features. After that, image processing is used to extract micro-Doppler spectrogram envelop and feature extraction. Classification on human activities with K-Nearest Neighbors (KNN) is introduced. Finally, this method is verified by five human activities using LFMCW radar developed in the laboratory.