一种基于稀疏贝叶斯分解的抗运动干扰心率提取算法

A Motion-resistant Heart Rate Extracting Algorithm Based on Sparse Bayesian Decomposition

目的提出一种基于光电容积脉搏波(photoplethsmography,PPG)的抗运动干扰心率检测算法。方法算法利用同步记录的一路加速度信号频谱调制离散傅里叶变换(discrete fourier transform,DFT)稀疏基,在此基上对两路PPG信号进行稀疏贝叶斯分解。由于谱系数的同一稀疏约束,此方法能够识别并去除PPG信号谱中运动干扰产生的谱峰。最后从五类特征点中搜索心率谱峰。结果实验基于12组PPG数据,算法结果平均绝对误差为1.99次/分。结论本文方法在基于PPG信号的心率检测方面效果良好,具有准确度高,抗干扰性强的特点。

Objective To propose a method for heart rate monitoring during physical activities using photoplethysmography（PPG）. Methods The spectrums of 2 channel PPG signals were estimated based on a Discrete Fourier Transform（DFT） sparse basis,which was modulated by the spectrum of one-channel simultaneously recorded acceleration signal. With a common sparsity constraint on the spectral coefficients,the method could easily identify and remove the spectral peaks of motion artifact（MA） in the PPG spectrum. Finally,the spectral peak corresponded to heart rate was searched from five kinds of feature points. Results Experiments were conducted on 12 groups of PPG signals. The average absolute estimation error was 1. 99 beats/min and the standard deviation was 2. 44 beats/min. Conclusion The method is of a good performance in PPG-based heart rate monitoring. The accuracy and the motion-resistant ability of this method are satisfactory.