基於光體積描記法(Photoplethysmography, PPG)的手腕心率監測，使用者在靜態中測量會有較高的準確性，但是在動態中測量則會有較大的誤差。由於PPG訊號會受到手腕動作所引起的偽影雜訊干擾，影響心率測量的準確度。為了改善這個在動態中測量心率的問題，本研究提出一個方法由，來達到抑制動作偽影以及提高心率估計準確度的目的。這方法由兩階段的策略組成。第一階段分析由PPG訊號與三軸加速度訊號所得出的頻譜，利用所提出的抑制動作偽影方法識別出PPG頻譜上的偽影區間並抑制其偽影干擾。第二階段基於短時間內心率的連續性，利用前一刻的心率縮小心率估計的搜尋範圍，提高動態中PPG心率估計的準確性。12位受測者的心率估計誤差平均值為2.57 BPM，有效減少運動過程中所引起的心率測量上的失真。;While the accuracy of heart rate monitoring using photoplethysmography (PPG) is quite acceptable when the person stays still, it could deteriorate during physical activities. The motion artifacts due to hand movements in the received PPG signal interfere with the heart rate detection. In this study, we propose a two-stage strategy to reject the motion artifacts and improve the accuracy of heart rate detection. At the first stage, the frequency spectrum of the received PPG signal is analyzed and the spectral components owing to the motion artifacts are identified and deleted by spectral subtraction. To detect the frequency spectrum of the motion signal, three-axis accelerometers are used. At the second stage, the heart rate is estimated on the basis of the spectral peaks of next moment and the tracking information of previous heart rates. This tracking estimation method has considerably improved the accuracy of PPG heart rate monitoring during physical activities. Experiments with 12 test subjects resulted in an average heart rate error of 2.57 BPM.
