研究目的
To develop an efficient algorithm for heart rate estimation using frequency spectrum analysis on photoplethysmographic (PPG) signals, enabling low-cost and real-time monitoring with minimal hardware resources.
研究成果
The proposed algorithm is efficient for real-time heart rate measurement in on-demand and continuous applications at rest conditions, achieving high accuracy with minimal hardware resources, making it suitable for low-cost embedded platforms; however, it is not effective during motion, indicating a need for further development in motion artifact removal.
研究不足
The algorithm does not produce good quality output when the subject is moving, as it is designed for rest conditions only; future work is needed to handle motion artifacts.
1：Experimental Design and Method Selection:
The algorithm involves pre-processing stages including windowing, sampling rate reduction, lowess smoothing, Hilbert transform, normalization, and FFT for frequency domain analysis to estimate heart rate based on maximum amplitude frequency components.
2：Sample Selection and Data Sources:
PPG signals collected from reflective type sensors on 100 subjects (75 white skin, 25 dark skin) in rest conditions, with data used for on-demand and continuous testing.
3：List of Experimental Equipment and Materials:
Reflective type PPG sensors with IR and Red LEDs, photo detector, cortex M4 platform, Keil uvision 4 IDE, snapdragon evaluation board for real-time testing.
4：Experimental Procedures and Operational Workflow:
Input PPG signal is windowed for 5 seconds with 80% overlap, decimated to reduce sampling rate, smoothed using lowess filter, processed through Hilbert transform and normalization, then FFT is applied to find frequency components for HR estimation, with SNR thresholding for output validation.
5：Data Analysis Methods:
Performance metrics include pass percentage, MAPE, MAE, RCF; comparisons made with other methods like SVD, EMD, ICA; statistical analysis using equations for HR calculation and SNR.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
