New insights on super-high resolution for video-based heart rate estimation with a semi-blind source separation...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computer Vision

Background:

• Remote photoplethysmography (rPPG) is a non-contact method for estimating heart rate (HR) from video.
• Low-resolution videos can yield rPPG signals, but quality may degrade due to limited skin pixels and camera noise.
• Super-high resolution's impact on rPPG HR estimation across different distances is not fully understood.

Purpose of the Study:

• To comprehensively investigate the benefits of super-high resolution for rPPG-based HR estimation.
• To evaluate the influence of shooting distance on rPPG measurement quality with varying resolutions.
• To introduce and test a novel semi-blind source separation (semi-BSS) rPPG method.

Main Methods:

• Development and testing of a new semi-blind source separation (semi-BSS) rPPG method.
• Evaluation using public UBFC-RPPG and self-collected video datasets.
• Comparative analysis against existing rPPG techniques under different resolution and distance conditions.

Main Results:

• The proposed semi-BSS method demonstrates superior performance compared to several existing rPPG techniques.
• Super-high resolution consistently and significantly improves rPPG signal quality when the shooting distance is 1.0m or greater.
• The study observed a notable enhancement in rPPG measurement quality with increased resolution at longer distances.

Conclusions:

• Super-high resolution offers a consistent and remarkable improvement in rPPG signal quality for HR estimation, particularly at distances of 1.0m and beyond.
• The choice of video resolution, tailored to the shooting distance, is critical for optimizing rPPG measurement accuracy.
• The novel semi-BSS method shows promise for robust rPPG-based HR monitoring.