Non-Contact Life Signal Extraction and Reconstruction Technique Based on MAE

Abstract

Non-contact life signal extraction could be used in natural disaster rescue, health care and other fields. As the energy of respiration and heartbeat signals is extremely weak in reality, they are usually submerged in noise and clutter. As a result, the traditional life signal extraction algorithms always fail at low signal-to-noise ratio (SNR) conditions. This paper proposes a novel life signal extraction and reconstruction algorithm based on MTI-Autocorrelation-EEMD (MAE) so as to enhance the accuracy and stability of life signal detection at low SNR condition. Taking advantage of its strong robustness with regard to noise, this technique utilizes a moving target indicator (MTI) algorithm to eliminate the interference of fixed object clutter on the echo signal after pulse compression. Combined with the autocorrelation algorithm with stable detection performance, the precise location of the human body micro-motion signal can be determined. Through the integration of an ensemble empirical mode decomposition (EEMD) algorithm, the phase of human body micro-motion signal is adaptively decomposed, overcoming the problem of mode mixing so that the signals of respiration and heartbeat can be reconstructed in the time domain according to the mode judgement criteria. Extensive real data experiments illustrate the effectiveness and robustness of the proposed algorithm under low SNR.

Keywords

• Non-contact life signal extraction and reconstruction
• ensemble empirical mode decomposition (EEMD)
• moving target indicator (MTI)
• autocorrelation detection
• respiration and heartbeat frequencies estimation