Real-time realizable mobile imaging photoplethysmography.

Area of Science:

• Biomedical Engineering
• Robotics
• Computer Vision

Background:

• Remote heart rate monitoring using photoplethysmography imaging (PPGI) sensors offers a non-contact alternative to traditional methods.
• Existing PPGI techniques often require complex algorithms or specific hardware setups.
• There is a need for autonomous and accurate remote heart rate estimation systems.

Purpose of the Study:

• To develop and evaluate a robot-mounted PPGI system for active and autonomous heart rate (HR) estimation.
• To propose a novel, efficient algorithm for real-time HR assessment using facial skin color analysis.
• To validate the algorithm's accuracy and processing speed against existing methods and datasets.

Main Methods:

• Mounted PPGI sensors on a robot for active and autonomous heart rate estimation (R-AAH).
• Developed a simplified algorithm for facial skin image extraction using HSV color space saturation (S) values.
• Selected facial pixels based on the most frequent S value for reliable HR assessment.
• Validated the algorithm on the UBFC-RPPG and BAMI-RPPG datasets.

Main Results:

• The proposed algorithm achieved a low average absolute error (AAE) of 0.71 bpm on the UBFC-RPPG dataset.
• The R-AAH method demonstrated high accuracy compared to existing algorithms.
• The algorithm processes data rapidly, with a processing time under 1 second (275 ms for an 8-s window).
• Validation on the BAMI-RPPG dataset yielded an AAE of 0.82 bpm.

Conclusions:

• The developed robot-based PPGI system enables accurate and autonomous remote heart rate estimation.
• The simplified algorithm offers an efficient and reliable method for HR assessment using facial imaging.
• This approach provides a promising non-contact solution for continuous HR monitoring.