Video-Based Pulse Rate Variability Measurement Using Periodic Variance Maximization and Adaptive Two-Window Peak...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Physiological Monitoring

Background:

• Remote photoplethysmography (rPPG) accurately measures Heart Rate (HR).
• Measuring Pulse Rate Variability (PRV) remotely is challenging due to noisy rPPG signals and lower temporal resolution compared to contact methods.
• Accurate PRV measurement is crucial for applications like remote stress and emotion recognition.

Purpose of the Study:

• To enhance non-contact PRV measurement accuracy using rPPG.
• To introduce and validate the Periodic Variance Maximization (PVM) method for rPPG signal extraction.
• To improve peak detection in rPPG signals for PRV analysis using an event-related Two-Window algorithm.

Main Methods:

• Utilized the Periodic Variance Maximization (PVM) method for rPPG signal extraction.
• Employed an event-related Two-Window algorithm for improved peak detection in temporal rPPG signals.
• Developed an adaptive parameter determination method for the Two-Window algorithm.
• Compared the proposed algorithm against the Slope Sum Function (SSF) and Local Maximum methods.

Main Results:

• Demonstrated the efficacy of the PVM and Two-Window algorithm for non-contact PRV measurement.
• The proposed adaptive Two-Window method successfully optimized parameters for PRV analysis.
• The developed algorithm outperformed existing methods, including SSF and Local Maximum, in accuracy based on contact-based ground truth.

Conclusions:

• The PVM method combined with the adaptive Two-Window algorithm represents a significant advancement in non-contact PRV measurement.
• This approach offers a viable solution for accurate PRV assessment in scenarios where contact-based measurements are impractical.
• The findings pave the way for more reliable remote physiological monitoring and stress/emotion detection.