Real-time brainwave-controlled interface using P300 component in EEG signal processing

摘要

Electroencephalography (EEG) is an emerging held of digital signal processing. EEG is electrical signals which are recorded by the sensors attached on the human scalp to detect human's brain activities along the scalp. EEG signal processing has been a technically challenging problem for researchers due to its extremely noisy nature compared to other different kinds of digital data such as voice or image. However, EEG signals also create promising research areas and applications for the human being. This paper aims at designing a brain-controlled interface using P300 component in EEG signal processing. The EEG signal's P300 component is an event related potential (ERP) component elicited by the human's brain in the process of decision making. We use the oddball paradigm presented by the P300-Speller to create the brain-computer interface for the subject to express their intention of thinking via the computer screen. The device used to record the brain signal is the Emotiv EPOC headset. We exploit the Bayesian linear discriminant analysis (BLDA) to classify the P300 signals. We conduct the experiments to help people express their thought in l-out-of-4-choice interfaces in real time. The choices, which are presented on the computer screen, could be anything from letters to images or specific symbols. The experimental results reveal the acceptable accuracy of correct classification to be 80 percent and the best bit rate of over 10 bits per minute.