Research On The Heart Rate Extraction Algorithm Of Compressive Sampled Photoplethysmography Signals

Cardiovascular disease has caused a serious threat to human health at present.Heart rate is a very important human parameter.Long-term and accurate heart rate testing can help people understand their physical condition at any time.For cardiovascular patients,early prevention and early treatment can be achieved.purpose.The pulse wave monitoring technology based on the photovolume method has been widely used in the daily monitoring of the cardiovascular state of the human body,but based on pulse signal sampling of the traditional Nyquist sampling theorem,the large amount of data generated will not only increase the overall power consumption of the system,It will also bring a greater burden in the process of transmission,storage and processing,resulting in a greater waste of system resources.Photoplethysmography-based detection technology is also very susceptible to motion artifacts,seriously affecting the accuracy of heart rate detection,under severe sports conditions,using conventional methods can not accurately obtain the heart rate value for the use of pulse signal to achieve long time Heart rate monitoring suffers from high power consumption and is susceptible to motion artifacts.This paper presents a method for extracting heart rate from compressed sampling pulse signal when there is motion interference.The main work is as follows:1)This paper describes the formation mechanism of the photoplethysmographic pulse signal and details the causes of motion artifacts.The sparse representation problem of pulse signals in different transform domains is analyzed,and a 0-1 sparse matrix which is easy to implement in hardware is designed.On this basis,the reconstruction performance of pulse signal under fast sparse Bayes algorithm is analyzed.2)For the heart rate extraction of the pulse signal with compressed sampling,the conventional method cannot be used in the time domain and frequency domain.In the time domain,a heart rate extraction method based on matched filtering is proposed.By establishing the time domain template of the PPG signal,the temporal heart rate is obtained by detecting the correlation degree between the template and the compressed observation signal.In the frequency domain,based on the least squares spectrum analysis method,the average value in each frequency domain window is estimated at a higher rate by using the sliding window idea.3).First,using Lomb-Scargle spectral analysis method to extract the spectrum from the compressed pulse signal and the uncompressed acceleration signal,then use the least squares spectrum subtraction to obtain the differential spectrum of the two signals,and finally perform the peak tracking to obtain the compressed sampling pulse signal.Heart rate estimate.The experimental results show that the accuracy of the heart rate estimation is improved compared to the existing methods at a compression rate of 25 times.The method reduces the amount of data collected,reduces the system resources occupied by data processing,storage and transmission,reduces the overall power consumption of the system,and has strong anti-interference and better heart rate estimation performance.