Méthodes d'analyse et de débruitage multicanaux à partir d'ondelettes pour améliorer la détection de potentiels évoqués sans moyennage : application aux interfaces cerveau-ordinateur

Abstract : Brain-Computer Interfaces (BCI) are control and communication systems which were initially developed for people with disabilities. The idea behind BCI is to translate the brain activity into commands for a computer application or other devices, such as a spelling system. The most popular technique to record brain signals is the electroencephalography (EEG), from which Event-Related Potentials (ERPs) can be detected and used in BCI systems. Despite the BCI popularity, it is generally difficult to work with brain signals, because the recordings contains also noise and artifacts, and because the brain components amplitudes are very small compared to the whole ongoing EEG activity. This thesis presents new techniques based on wavelet theory to improve BCI systems using signals' similarity. The first one denoises the signals in the wavelet domain simultaneously. The second one combines the information provided by the signals to localize the ERP in time by removing useless information
Document type :
Theses
Complete list of metadatas

Cited literature [144 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-02074919
Contributor : Memoires Ul <>
Submitted on : Thursday, March 21, 2019 - 9:15:54 AM
Last modification on : Saturday, March 23, 2019 - 1:24:23 AM
Long-term archiving on : Saturday, June 22, 2019 - 12:40:48 PM

File

DDOC_T_2013_0138_SAAVEDRA.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02074919, version 1

Collections

Citation

Carolina Véronica Saavedra Ruiz. Méthodes d'analyse et de débruitage multicanaux à partir d'ondelettes pour améliorer la détection de potentiels évoqués sans moyennage : application aux interfaces cerveau-ordinateur. Informatique [cs]. Université de Lorraine, 2013. Français. ⟨NNT : 2013LORR0138⟩. ⟨tel-02074919⟩

Share

Metrics

Record views

35

Files downloads

20