Skip to Main content Skip to Navigation

Diagnostic de la somnolence d'un opérateur: analyse automatique de signaux physiologiques

Abstract : In order to reduce the huge number of accidents, the LAAS-CNRS is interested since several years by determining the vigilance level on an operator by analysing his mode of driving. The last works realised in this field were concentrated on driving on motorway and on experimental conditions such as somnolence, as one origin of the hypo vigilance. In such study, the physiological measurements were served as reference of drowsiness level for behavioural measurements. My work, which concerns the automatic analysis of the physiological signals (EEG, EOG) in order to characterize the somnolence, is a part of European project SENSATION. This work is articulated in 2 parts. The first one is dedicated to the EEG analysis and starts with a presentation of analysed signals and the phenomena representative of somnolence, then the time frequency analyse methods: Fourier Transform, Wavelets decomposition and Hilbert Huang Transform. The last method was selected for the rest of study. We propose an empirical study of the accuracy for this technique and a method to increase this accuracy when the amplitude of the analysed signal is small. Then we present the developed algorithm and the first results obtained on real signals. The second part of the thesis is devoted to localisation and characterisation of blinks in EOG. Before presenting the developed algorithm, we describe the analysed signal and the suggested model for treating this signal. Then, we compare our results obtained by our algorithm with a semi-automatic analysis. Once the blink detection is validated, we compare the drowsiness level determined automatically with the visual analysis of the signals. The final results and the perspectives for improving these results is conclude this work.
Complete list of metadata

Cited literature [130 references]  Display  Hide  Download
Contributor : Arlette Evrard Connect in order to contact the contributor
Submitted on : Friday, December 14, 2007 - 1:58:29 PM
Last modification on : Tuesday, October 19, 2021 - 11:16:24 PM
Long-term archiving on: : Monday, April 12, 2010 - 7:30:54 AM


  • HAL Id : tel-00197161, version 1


Hassan Sharabaty. Diagnostic de la somnolence d'un opérateur: analyse automatique de signaux physiologiques. Micro et nanotechnologies/Microélectronique. Université Paul Sabatier - Toulouse III, 2007. Français. ⟨tel-00197161⟩



Record views


Files downloads