Noise reduction of EEG signal based on head movement estimation by using frontal viewing camera

Jae Won Bang, Jong Suk Choi, Eui Chul Lee, Kang Ryoung Park, Mincheol Whang

Research output: Contribution to journalArticlepeer-review

5 Scopus citations

Abstract

Recently, the use of the brain computer interface (BCI), which enables interaction with a computer via a user's electroencephalogram (EEG) signal, has been researched for several applications. Examples of such applications include control of mobile devices, analysis of the driving habits and drowsiness of drivers, diagnosis of brain diseases such as epileptic seizure, and analysis of eye fatigue on display devices. In general, several electrodes should be attached to the surface of the head in order to measure the EEG signal, but noise occurs due to the movements of the attached electrodes by the user's head movements. In order to solve this problem, we propose a new method for correctly obtaining robust EEG data from head movements. This research is novel in the following three ways. First, a user's head movements are estimated by using a frontal viewing camera attached to a wearable device. Second, in order to estimate the accurate direction and amount of head movement with successive images by the frontal viewing camera, motion vectors between the images are obtained by the Lucas-Kanade-Tomasi (LKT) method. Third, whether or not the user's head movement occurs is determined by using a user-dependent support vector machine (SVM) with two features based on the motion vectors. The experimental results showed that the proposed method could measure the EEG signals robust to the user's head movements.

Original languageEnglish
Pages (from-to)1241-1246
Number of pages6
JournalSensor Letters
Volume10
Issue number5-6
DOIs
StatePublished - May 2012

Keywords

  • Brain Computer Interface
  • Electroencephalogram
  • Lucas-Kanade-Tomasi
  • User-Dependent Support Vector Machine

Fingerprint

Dive into the research topics of 'Noise reduction of EEG signal based on head movement estimation by using frontal viewing camera'. Together they form a unique fingerprint.

Cite this