A non-contact vital sign detection method based on modulated continuous wave（ FMCW） radar is proposed for multi-target scene. Moving target indication（ MTI）

exponential weighted average filtering and recursive average filtering are used to remove the static clutter and background noise contained in the echo signal

which greatly improves the signal-to-noise ratio; CA-CFAR adaptive detection threshold and minimum distance threshold are used to detect multiple targets. Nearest neighbor correlation algorithm and Kalman filtering are combined to achieve multi-target tracking. By processing the extracted phase signals of multiple targets

the respiratory and heartbeat signal are obtained. The final heart rate and respiratory rate are obtained by combining fast Fourier transform（ FFT） and autocorrelation method. Experiments show that the proposed method can accurately and stably realize the real-time detection of multi-target vital signs.