• Aucun résultat trouvé

Test of a new drowsiness monitoring system based on images of the eye

N/A
N/A
Protected

Academic year: 2021

Partager "Test of a new drowsiness monitoring system based on images of the eye"

Copied!
1
0
0

Texte intégral

(1)

TEST OF A NEW DROWSINESS MONITORING SYSTEM BASED ON IMAGES OF THE EYE

C François1, T Langohr1, T Hoyoux1, J Wertz1, J Verly1

1Laboratory for Signal and Image Exploitation, Department of Electrical Engineering and

Computer Science, University of Liège, Liège, Belgium Abstract Text:

The involvement of drowsiness in a large number of accidents, whether in private or professional activities, makes its detection a major societal issue to improve public health and safety.

Drowsiness is the intermediate state between wakefulness and sleep. It is characterized by impairments of performance, e.g. slow reaction time, loss of vigilance, and deficits in information processing; which can be very dangerous, especially in the case of driving. Moreover, all persons are affected by drowsiness but some people have a higher propensity than normal to drowsiness. Indeed, 6 to 11% of the population suffer from excessive daytime sleepiness. There is thus an obvious need for a device able to detect drowsiness in an objective way, and to avoid tragic accidents.

We have thus developed a new, automatic, real-time drowsiness monitoring system based on the physiological state of a subject. This system uses ocular parameters extracted from images of an eye to determine a level of drowsiness on a numerical scale. Oculography is indeed recognized as one of the most significant and practical technique to characterize drowsiness. In order to “validate” our system, we compared, for a number of subjects, the level of drowsiness determined by our system to two references: (1) the level of drowsiness obtained by analyzing polysomnographic signals, and (2) the performance of these subjects in the accomplishment of a task. We thus realized an experiment in which 25 subjects were asked to perform several reaction time tests in different sleep-deprivation conditions. Results show that our system is well correlated with both references. Moreover, it has the advantage of being non-invasive, usable in any condition, and of requiring no intervention from the subject. It has thus significant potential for reliably quantifying the level of drowsiness of subjects accomplishing a task and ultimately for preventing drowsiness-related accidents.

Références

Documents relatifs

A regular MCA was applied to this data, as well as a sparse MCA with the constraints that each dimension be based on only one group of variables (i.e., a unique categorical

( ﺫﻴﻔﻨﺘ ﺭﺎﻁﺇ ﻲﻓ ﺔﻠﻴﺴﻭﻟﺍ ﻩﺫﻫ لﺎﻤﻌﺘﺴﺇ ﻯﺩﻤ ﻰﻠﻋ ﻑﻭﻗﻭﻟﺍ ﻱﺭﻭﺭﻀﻟﺍ ﻥﻤ ﻪﻨﺈﻓ ﻲﻟﺎﺘﻟﺎﺒﻭ ﻕﻭﺴﻟﺍ ﺩﺎﺼﺘﻗﺍ ﻯﻭﺤﻨ لﺎﻘﺘﻨﻹﺍ ﻭ ﺔﻴﺩﺎﺼﺘﻗﻹﺍ ﺕﺎﺤﻼﺼﻹﺍ. ﺔﺴﺍﺭﺩﻟﺍ ﺭﺎﻁﺍ ﺩﻴﺩﺤﺘ :

The point with detecting α bursts in EEG signal is the difficulty to define a common threshold for all drivers because of the large inter-individual differences

archives, véritable mémoire du journal et donc d'une certaine vision de la société, acquièrent une dynamique nouvelle et favorisent un contact plus étroit avec le (cyber)lectorat qui

Based on this idea, we define the Rascal 1 technique that uses both pieces of information – the joint likelihood ratio and the joint score – simultaneously to train an estimator ˆ

According to this study, patients previously treated with Infliximab with high levels of ATIs have a lower response rate to Adalimumab than patients with low levels of

« Les élèves les plus “faibles” ne parviennent pas à mémoriser de façon durable les correspondances entre graphies et sons, les combinaisons entre lettres ou même à

Assaf and Samuel-Cahn (1996) prove that for any discrete memoryless threshold rule defined by a sequence which is not monotone increasing, there exists a rule determined by a