A method for eliminating motion noise of pulse wave signal

Abstract

The invention provides a method for eliminating pulse wave signal motion noise. The method includes the steps that 1, a noise-containing PPG signal and a noise-containing acceleration signal are obtained; 2, a parallel adaptive filter is used, the noise-containing PPG signal is filtered on the basis of the acceleration signal, and a relatively pure PPG signal is obtained; 3, Fourier transformationis carried out on the acceleration signal and the relatively pure PPG signal, and an acceleration frequency spectrum and a pulse wave frequency spectrum are obtained; 4, spectral subtraction is carried out on the pulse wave frequency spectrum on the basis of the acceleration frequency spectrum to obtain a sparse PPG spectrum; 5, sparse signal reconstruction is performed on the sparse PPG spectrumto obtain a pure PPG signal. The method can eliminate the small-amplitude motion noise, and a pure pulse wave signal can still be obtained under scenes that the motion amplitude ratio is large. The method has certain robustness, is suitable for wearable physiological index monitoring equipment and has high practicality.

Description

technical field [0001] The invention relates to the field of digital signal processing, in particular to the field of digital signal processing in biomedical engineering. Background technique [0002] Photoplethysmographic signal (PPG) is the light intensity signal of reflected light or refracted light obtained after the light wave irradiates the skin. The fluctuation of this signal is based on the pulse of the artery. The change causes the light absorption to change, while the light absorption of skin, muscle, venous blood and other tissues remains constant. According to the Lambert-Beer law, the change of arterial blood volume is estimated by monitoring the change of reflected or refracted light intensity, and the pulse wave signal is obtained, and then the physiological characteristics such as heart rate, blood pressure, and blood sugar are obtained through the pulse wave signal. [0003] At present, the mainstream device for pulse wave collection on the market is a wris...

AI Technical Summary

Problems solved by technology

In addition, it is necessary to estimate and judge the number of layers that the noise may be distributed in advance. The uncertainty of the motion in the actual process cannot be adaptively dealt with, and it cannot handle the situation that the real pulse wave signal is similar in frequency to the motion noise and is in the same layer.

Therefore, wavelet transform denoising is only suitable for small movements, such as arm swings and other forms of movement, not suitable for sports scenes such as running

[0008] 4) Empirical Mode Decomposition (EMD), this method is similar to wavelet analysis, it is automatically decomposed into IMFs of several frequency bands according to the nature of the signal itself, and there is also the problem of wavelet transform denoising

[0011] Most of the existing pulse wave motion noise reduction technologies can only eliminate small motions, but there is no better way to deal with the noise caused by strenuous motion

Moreover, most motion noise cancellation technologies have relatively high computational complexity and cannot be processed in real time, which makes them useless in many scenarios.

Images