Drowsiness Detection System Based on PERCLOS and Facial Physiological Signal

Area of Science:

• Biomedical Engineering
• Transportation Safety
• Human-Computer Interaction

Background:

• Fatigue-related accidents are a significant public safety concern.
• Current driver monitoring systems often rely on intrusive methods or lack robustness.
• Assessing driver alertness through physiological signals and eye-tracking is crucial for accident prevention.

Purpose of the Study:

• To develop an accurate and robust non-contact drowsiness detection system for drivers.
• To integrate heart rate variability (HRV) analysis with eyelid closure monitoring (PERCLOS) for improved fatigue assessment.
• To validate the system's performance against electroencephalography (EEG) signals.

Main Methods:

• Utilized photoplethysmographic imaging (PPGI) to derive the LF/HF ratio from HRV, indicating sympathetic/parasympathetic nervous system balance.
• Employed a near-infrared webcam for non-contact, low-light facial image acquisition to calculate PERCLOS.
• Developed an algorithm integrating HRV and PERCLOS for comprehensive drowsiness judgment.

Main Results:

• The system achieved high accuracy (92.5%) with a sensitivity of 88.9% and specificity of 93.5%.
• Non-contact measurement using a near-infrared webcam proved effective even in dark environments.
• The combined approach demonstrated improved robustness compared to individual methods, validated by EEG comparison.

Conclusions:

• The proposed integrated system offers a reliable and non-intrusive method for detecting driver drowsiness.
• This technology has the potential to significantly reduce fatigue-related road accidents.
• Further research can explore real-world implementation and integration into vehicle safety systems.