Facial Privacy Protection for Remote Photoplethysmography

Area of Science:

• Biomedical Engineering
• Computer Vision
• Signal Processing

Background:

• Remote photoplethysmography (rPPG) enables contactless health monitoring using skin videos.
• Facial rPPG measurements raise privacy concerns due to sensitive biometric data.
• Existing privacy and compression methods degrade rPPG signal quality.

Purpose of the Study:

• To develop a face anonymization technique for rPPG videos.
• To preserve physiological information during anonymization.
• To enhance video compression efficiency for rPPG applications.

Main Methods:

• A plug-and-play face anonymization module using spatial pixel redistribution.
• Eliminating identifiable biometric features while retaining physiological data.
• Macroblock reassembly strategy based on chromaticity clustering for compression.

Main Results:

• The proposed method effectively anonymizes facial features in rPPG videos.
• Physiological information crucial for rPPG measurements is preserved.
• Anonymized videos show improved facilitation of video compression.

Conclusions:

• The developed face anonymization module addresses critical privacy and compression challenges in rPPG.
• This technique supports the advancement of telemedicine and large-scale medical dataset creation.
• The method offers a practical solution for privacy-preserving rPPG applications.