本篇論文主要為穩態視覺誘發電位(steady state visual evoked potential, SSVEP)腦波人機介面(brain computer interface, BCI)的相關研究，藉由人體注視閃光刺激會產生誘發訊號的特性，我們提供四個固定頻率但不同相位編碼的LED閃爍光源，利用梳狀濾波器(comb filter)在數位濾波器容易實現，且可去除不需要的雜訊干擾，保留主頻率及其諧波的特性，並由事件相關性(correlation)來判斷閃光的相位，分辨出正確的指令；整個系統利用自製的類比腦波擷取放大器搭配單晶片微控制器(micro control unit, MCU)設計完成，目的在建立一個低成本、體積小且能快速辨識的腦波人機介面系統。本系統目前可讓使用者進行四個按鍵的輸入，其平均準確率為91.64%，平均資料傳輸速度(information transfer ratio, ITR)可達到31.02(bits/min)。This thesis mainly designs as a brain computer interface (BCI) system for electroencephalogram (EEG) of steady state visual evoked potential (SSVEP). Since users gaze at different spatially separated flash channels (FCs) in order to induce visual evoked signals, the BCI provides four fixed frequency but different phase encoding in the flickering source. The Algorithm uses comb filter in the digital filter easy to implement, keep main frequency and harmonic, and can reduce noise. In order to recognize the command mapping to the gazed FC can be sent out to achieve control purposes, the current design uses event correlation to achieve identify distinct flickering sequences among different FCs. The implementation method is designed the analog EEG capture amplifier and micro control unit(MCU), in order to establish a low cost, small size and fast to recognize BCI system. In this thesis, we have built an four-FC system. The command information transfer ratio(ITR) and detected accuracy are 31.02 bits/min and 91.64%, respectively.
