Practical Classification Techniques In The Brain - Computer Interface System

In recent years, it is possible to research brain-computer interface (BCI) with the rapid development of computer technology and the wider application of signal processing technology. BCI is a direct data communication channel that transmits information from a human brain to a computer or other electronic equipments, instead of relying on brain activity transmitted through the pathways of peripheral nerves and muscles. It can provide a new way of communication and control for paralysis patients, especially who lost the basic physical movements but thinking. So, it is being received increasing attention.At present, the electroencephalogram (EEG) machine for BCI research is mostly special, which has many advantages such as convenient experimental design, rapid data extraction and so on, but very expensive. On the other hand, the clinical machine can not be used in BCI research because of the lack of corresponding software platform to support though it is relatively cheap. So, in this paper, the author used first Microsoft Visual C++ 6.0 to design a software system named sdund. It can collect event-related data from USB interface for graphics to show in real-time manner, the final form of data files to data saving. It adopt Microsoft SQL Server 2000 database for saving information, designed the forefinger pressing keyboard experiment.BCI research in this paper belongs to classification of single-trial EEG during finger movements. This technology has become an attractive topic in BCI research due to its advantages such as simple experimental approach, short response time and high classification accuracy. This paper presents an algorithm based on bereitschaftspotential (BP) and event-related desynchronization (ERD), which used support vector machine (SVM) as classifier for classifying single -trial EEG during voluntary forefinger pressing keyboard. After we got EEG data from above sdund system, the multichannel EEG is preprocessed by low-pass and band-pass filters because BP occurs in the low frequency band and ERD in the middle frequency band, then the fixed time window is used to filter...