Anasayfa







Sempozyum: 6th Internatıonal Computer and Instructıonal Technologies Symposium
Electroencephalogram Based Human-Computer Interfaces
Onder AYDEMİR
Karadeniz Technical University

Temel KAYIKÇIOGLU
Karadeniz Technical University
Bildiri Özeti:
Anahtar Sözcükler:


Sempozyum: 6th Internatıonal Computer and Instructıonal Technologies Symposium
Electroencephalogram Based Human-Computer Interfaces
Onder AYDEMİR
Karadeniz Technical University

Temel KAYIKÇIOGLU
Karadeniz Technical University
Abstract :

The key idea behind electroencephalogram based human computer interface (EB-HCI) system is to allow paralyzed subjects to interact the external world without using their muscles. Electroencephalogram (EEG) is the most used potential for HCI designs, mainly due to its fine temporal resolution, non-invasiveness, easy implementation and low set-up costs. EB-HCI systems are analyzed electrical brain activity recorded from electrodes placed the subject’s scalp and extract feature(s) to determine the intents of the user. Then, translate them into the control signals that are used to control external devices (e.g., a robotic arm, a wheelchair). Current EBHCI systems are generally performed in five main steps. These are respectively:

 
i. Signal Acquisition: EB-HCI systems begin with the signal acquisition. In this step the EEG signals are captured and digitized for further analysis. The EEG signals are sampled generally with 256 Hz. On the other hand, a 50 Hz notch filter is used to eliminate line noise. International 10- 20 System is the most widely used method to describe the placement of electrodes at specific intervals along the head. Each electrode site has a letter to identify the lobe, along with a number or another letter to identify the hemispheric location.


ii. Preprocessing: Because of EEG and electrocorticogram (ECoG) signals are not stationary and contaminated
with various artifacts, such as electromyogram (EMG) and electrooculogram (EOG) a preprocessing stage is applied to prepare the signals in a suitable form for further steps. An appropriate preprocessing algorithm in EB-HCI system has great importance with respect to extract the best features to discriminate different mental tasks.


iii. Feature Extraction: Feature extraction is necessary for represent input signals in a reduced feature space
and identifying discriminative information for every kind of signals that have been recorded.


iv. Classification: The main task of this step is to categorize the signals between a fixed set of classes by taking feature vectors into account. Selection of discriminative feature(s) and the most appropriate classifier are very critical problem, because they are directly aff ects recognition rate of EB-HCI system. Hence, the mostly used feature extraction (Fourier transform, wavelet transform etc.) and classification (k-nearest neighbor and support vector machines and linear discriminant analysis) methods are discussed by considering their advantages and critical points in the paper.

 
v. Device Control Interface: This step translates the categorized signals into the control commands that are
used to control external devices.
In this paper, those five steps are introduced in more detail. Moreover, this paper provides a review of the
EB-HCI technology for researchers who are interested in that field.

Keywords : human computer interface, electroencephalogram, feature extraction, classification
Dokümanlar / Documents
0506 891 83 48 | 0541 434 54 24
0312 425 90 90 | 0549 548 06 30
Bu web sitesinde yer alan bütün görsel ve yazılı materyallerin telif hakları Pegem Akademi Yayıncılık'a aittir.
Her hakkı saklıdır. 2005-2018 © Pegem Akademi Yayıncılık.
.