Long Distance Vital Signs Monitoring with Person Identification for Smart Home Solutions.

Area of Science:

• Biomedical Engineering
• Computer Vision
• Signal Processing

Background:

• Remote vital sign monitoring is crucial for healthcare, especially for individuals requiring constant supervision.
• Existing imaging photoplethysmography (iPPG) systems are limited to short distances (<1m), hindering real-world applications.
• Long-distance, non-contact vital sign measurement offers enhanced convenience and reliability for diverse user groups.

Purpose of the Study:

• To investigate the feasibility of remote pulse estimation at distances exceeding 5 meters using visible light.
• To integrate deep learning for robust person tracking and identification in remote monitoring scenarios.
• To enable personalized vital sign pattern analysis and multi-person monitoring from a single video stream.

Main Methods:

• Utilized a webcam for acquiring visible light sequences at a distance of 6 meters.
• Implemented advanced image pre-processing algorithms to enhance signal quality.
• Integrated a deep learning model for person tracking and identification, even without visible facial features.

Main Results:

• Successfully extracted pulse rate from visible light sequences acquired at 6 meters.
• Achieved high accuracy with a Root Mean Square Error (RMSE) of less than 2.8 beats per minute.
• Demonstrated the capability for personalized vital sign monitoring and multi-person tracking.

Conclusions:

• Remote pulse estimation via visible light imaging is feasible at distances up to 6 meters.
• Deep learning integration enhances the robustness and applicability of iPPG systems for long-distance monitoring.
• This technology holds significant potential for non-contact, continuous vital sign surveillance in various healthcare settings.