Driving Fatigue Classification Based on Fusion Entropy Analysis Combining EOG and EEG

Wang, Hongtao; Wu, Cong; Li, Ting; He, Yuebang; Chen, Peng; Bezerianos, Anastasios · 2019 · Crossref

DOI: 10.1109/access.2019.2915533

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the critical issue of driving fatigue, a major contributor to traffic accidents, by developing a robust classification system that integrates electroencephalogram (EEG) and electrooculogram (EOG) signals. While previous methods using power spectrum density and sample entropy showed promise, they lacked sufficient accuracy for reliable real-time detection. To overcome this, the authors propose a fusion entropy analysis method that leverages the complementary information provided by EEG, which reflects internal cognitive states, and EOG, which captures external subconscious behaviors like eye movements. The experimental design involved twenty-two healthy participants who engaged in a 90-minute simulated driving task using a virtual reality system equipped with a racing wheel and three large screens. Data were collected using a wireless dry EEG headset with twenty-four electrodes and EOG electrodes placed around the eyes. Signal preprocessing included spatial filtering, detrending, and discrete wavelet transform to decompose EEG signals into delta, theta, alpha, and beta sub-bands. Feature extraction focused on five entropy metrics: sample entropy from vertical and horizontal EOG, and approximate entropy, sample entropy, and spectral entropy from the EEG sub-bands. These features were fused using canonical correlation analysis to maximize the correlation between EEG and EOG modalities while maintaining independence. The resulting six-dimensional feature vector was classified using a relevant vector machine (RVM) with a radial basis function kernel. The results demonstrated that the proposed fusion entropy method significantly outperformed classifications based on single sub-bands or individual entropy types. The system achieved an average classification accuracy of 99.1% ± 1.2% in distinguishing between alert and fatigued states. Statistical analysis confirmed that the fused features were superior to those derived from individual frequency bands, with significant differences observed between the fusion approach and sub-band-only methods (p < 0.001). The study also validated the induced fatigue through NASA Task Load Index questionnaires, which reported high mental demand and frustration levels among participants. The significance of this work lies in its demonstration that combining EEG and EOG via entropy analysis provides a highly accurate and effective indicator for driving fatigue detection. By utilizing wireless dry electrodes and a realistic simulated environment, the method offers a practical alternative for real-time monitoring systems. The high accuracy achieved suggests that multi-modal fusion can enhance the robustness of fatigue detection algorithms, potentially contributing to safer driving conditions and reduced accident rates.

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 qwen3.6-27b-nvidia on 2026-08-10; verification: verified_with_issues.

Topics

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

• drowsiness detection algorithms
• drowsiness

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