In the past decade, many researchers have dedicated their efforts to exploring brain computer interface (BCI)\ntechnology. With a growing number of investigations into the BCI and its related areas, BCI systems nowadays are not\nonly developed for the disabled but also for the normal and healthy people. In this paper, a signal-processing-based\ntechnique with its applications into the development of automated character recognition is introduced. The task of the\npattern recognition to such a BCI design problem was mainly accomplished based on the detection of P300 evoked\npotentials. We approached this detection problem by employing a template-matching-based method to extract the\nmorphological information from EEG signals first, and then applying a linear discriminant function (LDF) to the features\nselected for pattern classification. The entire detection process was further implemented on an existing BCI system\nplatform, called the BCI2000 system. The algorithm performance was evaluated using an existing reliable database\nprovided by BCI Competition 2003. Numerical experimental results produced by the database indicated that the\nproposed algorithm actually achieved 100% character recognition accuracy.
Loading....