Analysis of Driver Drowsiness Detection using EEG and EOG

K Firke, Omprakash; Manish Jain, Dr · 2018 · Crossref

DOI: 10.14419/ijet.v7i2.17.11557

archive: archived pipeline: cataloged

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of automatic driver drowsiness detection, a significant factor in approximately 20% of traffic accidents. The authors propose a system that integrates physiological signals from Electroencephalography (EEG) and Electrooculography (EOG) to improve reliability over camera-based visual cues alone. The motivation stems from the complementarity of brain and ocular activities, which are the primary indicators used by physicians to assess vigilance. The study aims to develop a detection algorithm that identifies drowsiness by analyzing specific EEG frequency bands and EOG blink patterns without requiring subject-specific training data. The methodology focuses on analyzing EEG signals within the 1–30 Hz range, specifically targeting the Alpha (8–12 Hz) and Theta (4–8 Hz) bands, which are characteristic of drowsiness and inactivity. The paper reviews existing detection systems, comparing learning-based techniques (such as neural networks and Hidden Markov Models) with non-learning approaches. The authors advocate for a threshold-based method using Short Time Fourier Transform (STFT) to analyze frequency content, as it is suitable for real-time applications and does not require extensive training datasets. For EOG, the system monitors blink duration and frequency, utilizing the amplitude-velocity ratio to detect vigilance decline. The approach is validated against established scales like the Karolinska Drowsiness Scale and the Objective Sleepiness Scale (OSS), which correlate EEG activity levels with eye movement speeds. Key findings indicate that while learning-based methods like neural networks can achieve high detection rates (up to 92% in some studies), they often suffer from poor generalization across different drivers and require large amounts of training data. In contrast, the proposed non-learning method offers a significant advantage by using a fixed threshold for detection, making it driver-independent and easier to implement. The paper highlights that STFT is preferred over Wavelet Transform for real-time processing. Furthermore, the analysis of EOG shows that blink detection via video or electrodes is effective, but video-based systems face challenges with lighting conditions and image processing complexity. The integration of EEG and EOG provides a more robust framework than either signal alone, as confirmed by the OSS criteria which link specific EEG wave appearances to eye movement behaviors. The significance of this work lies in the development of a portable, driver-independent drowsiness detection system. By avoiding the need for individual calibration or large training datasets, the proposed approach is more practical for widespread adoption in portable devices. The paper concludes that combining cerebral and ocular data enhances the reliability of drowsiness detection, addressing the limitations of current commercial systems that rely solely on visual cues or vehicle behavior. This multi-modal physiological approach offers a viable path for creating effective, real-time safety systems that can prevent accidents caused by fatigue.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• drowsiness
• microsleep
• gaze based attention detection
• vigilance
• workload measurement

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data
• Methodological Resource: tool software, validation psychometrics