Super-resolution spectral estimation in short-time non-contact vital sign measurement

Non-contact techniques for measuring vital signs attract great interest due to the benefits shown in medical monitoring, military application, etc. However, the presence of respiration harmonics caused by nonlinear phase modulation will result in performance degradation. Suffering from smearing and leakage problems, conventional discrete Fourier transform (DFT) based methods cannot distinguish the heartbeat component from closely located respiration harmonics in frequency domain, especially in short-time processing. In this paper, the theory of sparse reconstruction is merged with an extended harmonic model of vital signals, aiming at achieving a super-resolution spectral estimation of vital signals by additionally exploiting the inherent sparse prior information. Both simulated and experimental results show that the proposed algorithm has superior performance to DFT-based methods and the recently applied multiple signal classification algorithm, and the required processing window length has been shortened to 5.12 s.