| Summary: | 碩士 === 國立臺北科技大學 === 機電整合研究所 === 104 === This paper uses brain computer interface(BCI) and EEG signal acquisition system. The system is designed by the Neuroscan company. We will design a panel with Visual C# and this panel is designed by self. This experiment has ten subjects. We will start the panel when subjects are ready. This panel has preparation time of three seconds after the panel be started. There are three stages of the experiment. First stage is the first relaxation. This stage has thirty seconds. Second stage is test stage. This stage will appear a gray figure in four seconds to ten seconds by random. This gray figure will maintain 250 ms and appears one hundred times. There are two types of image contains a horizontal chart and a straight chart. The horizontal chart appears in the sixth, twenty-second, thirty-second, thirty-fourth, fortieth, fiftieth, ninety-first, ninety-second, and the ninth nineteen. The remaining number of figures is the straight chart. In the test stage, these subjects must press the button when the chart appears in the windows After the test stage is end, the third stage is the second relaxation. This stage has three minutes. These subjects easily watch the windows in the first and second relaxation. In this paper, this three stages called one round. The experiment has three rounds.
In this paper, These two types of label is attention label and non-attention label. The feature has band power(BP), common spatial pattern(CSP), and filter banks common spatial pattern(FBCSP). The classification has K-nearest neighbor(K-NN), liner discriminant analysis(LDA), support vector machine(SVM), and support vector data description(SVDD). The result will be appeared in the error rate. The result in the FBCSP of feature and SVM of classification is 23.99% in the error rate. The study will select three electrodes in the feature and the classification.
|
|---|
