Non-contact Detection of Vital Sign Based on ICCEMDAN-SpEn Combined with Wavelet Threshold Denoising

The detection of vital signs, especially respiratory rate and heartbeat rate, is of great significance for clinical diagnosis, daily health assessment and disaster rescue. In recent years, millimeter wave based non-contact vital sign detection has become a research hotspot. Aiming to overcome the noise influence of the target signal on the detection accuracy, one improved ICEEMDAN based on sample entropy (ICCEMDAN-SpEn) combined with wavelet threshold denoising is proposed. Firstly, the adaptive noise extraction ability of ICEEMDAN is analyzed to isolate all intrinsic mode function (IMF) components of the noisy target signals, and the sample entropy is introduced as the threshold to select the main IMFs; then wavelet threshold denoising is proposed to relieve the noise adaptively, which can better deal with the nonlinear and non-stationary signal; finally, IMFs after wavelet threshold denoising are reconstructed to attain target signals. The experimental results show that the proposed method can effectively relieve the noise contained in the vital sign signals and estimate the respiratory rate and heartbeat rate of human body accurately.