Page 1 of 1

P300 based character recognition

Posted: 27 Sep 2015, 01:55
by skundu90
Is it possible to classify the character without knowing the row-column information?

p300 signal is generated when human are looking at flashing light. From the EEG signal correctly detection of p300 is important. After detection of p300 our job is to find what time this p300 signal generated and at that time instant which row or column flashed. From this row-column combination we can get the original character.

Now if I don't have the row-column information. Is this possible to detect the character from EEG signal?  

Best regards
Sourav Kundu
514ec1002@nitrkl.ac.in
Research scholar
India

Re: P300 based character recognition

Posted: 29 Sep 2015, 11:49
by pbrunner
Skundu90,

there are two pieces of information that you need to detect the intended character.

1) When was a stimulus delivered, i.e., the time when a row or column is intensified.
2) Which stimulus was delivered, i.e., the identity of the row and column.

In theory you could detect when a target stimulus was presented from the EEG data by using a matched filter, however without knowing the identity of the target stimulus you cannot determine which character the subject intends to spell.


Regards, Peter