| Autor: |
Kiashari, Serajeddin Ebrahimian Hadi, Nahvi, Ali, Bakhoda, Hamidreza, Homayounfard, Amirhossein, Tashakori, Masoumeh |
| Předmět: |
|
| Zdroj: |
Multimedia Tools & Applications; Jul2020, Vol. 79 Issue 25/26, p17793-17815, 23p |
| Abstrakt: |
In this paper, a new non-intrusive driver drowsiness detection method is introduced based on respiration analysis using facial thermal imaging. Drowsiness is the cause of many driving accidents all over the world. Drivers' respiration system undergoes significant changes from wakefulness to drowsiness and can be used to detect drowsiness. Current respiration measurement methods are intrusive and uncomfortable making respiration the least measured vital sign during driving. In this paper, a new method is presented based on facial thermal imaging to analyze drivers' respiration signal non-intrusively. Thirty subjects are tested in a car simulator. They are fully awake at the beginning and experience drowsiness during the tests. The mean and the standard deviation of the respiration rate and the inspiration-to-expiration time ratio are extracted from the subjects' respiration signal. To detect drowsiness, the Support Vector Machine (SVM) and the K-Nearest Neighbor (KNN) classifiers are used. The Observer Rating of Drowsiness method is used for scoring the drowsiness level and validating the proposed method. The performance and the results of both methods are presented and compared. The results indicate that drowsiness can be detected with the accuracy of 90%, sensitivity of 92%, specificity of 85%, and precision of 91%. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink