UTHM Institutional Repository

Classification of eeg signals for human computer interface (hci) application

Mohd Azali, Noor Nadiah (2015) Classification of eeg signals for human computer interface (hci) application. Masters thesis, Universiti Tun Hussein Onn Malaysia.

[img]
Preview
PDF
NOOR_NADIAH_BT_MOHD_AZALI_24.pdf

Download (1MB)

Abstract

Brain Computer Interface (BCI) is one of the alternatives available in situation when all other typical interface such as joystick is not an option. This situation is generally true for users with severe motor impairment such as spinal injury who are unable to control wheelchair. In this research, method to classify EEG signals for controlling wheelchair for severe impairment users is proposed. The proposed system will be using a low cost consumer grade device, Neurosky Mindwave Mobile, to safely measured and acquired EEG data. Two types of model are proposed, the first one is based on visualizing colour model, and the other one is imagining doing motor task. Colours chosen are cyan, black, green and yellow as this colour are proven to generate high brain activity. For mental task, subjects are required to imagine doing motor task such as running, kicking, juggling, and signing a song. Data acquired will then go through simplest pre-processing stage to obtain signal contain enough information for classification. Classification implemented using linear classifier, Support Vector Machine as EEG brainwave is presumed to be linear. Results by trying different combination of task were analyzed to deduct the best way to classify direction which might work for controlling wheelchair.

Item Type: Thesis (Masters)
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101-5865 Telecommunication. Telegraph.
Depositing User: Mrs Hasliza Hamdan
Date Deposited: 04 May 2016 09:51
Last Modified: 04 May 2016 09:51
URI: http://eprints.uthm.edu.my/id/eprint/7796
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year