Abstract: Aiming at the problem of strong harmonic interference in vital signs detection based on Doppler radar, an improved method based on continuous wavelet transform (CWT) feature extraction is proposed. First, multiple sets of data sets with different time lengths are selected for the feature signal, and then CWT is performed in the wavelet domain. Respiratory signals, respiratory harmonics, and heartbeat signals are identified based on whether or not the spectral peak position is independent of the length of the data. On this basis, the heart rate (HR) is extracted and the frequency resolution is analyzed. Experimental result shows that the heartbeat signal can be quickly detected by using the method proposed in this paper using data from 4 s to 5 s. When the frequency resolution requirement is 0. 1 Hz, the multi-data HR extraction method has relatively satisfactory results in effectively eliminating harmonics and quickly identifying the heartbeat signal.