查詢結果分析
來源資料
頁籤選單縮合
題 名 | 想像幾何左右旋轉與左右手動之辨識率比較=Comparison of Classification between Imaging Hand Movements and Imaging Geometry Rotations |
---|---|
作 者 | 張菀珍; 葉榮木; 蔡俊明; 劉昀松; | 書刊名 | 資訊科學應用期刊 |
卷 期 | 6:1 2010.06[民99.06] |
頁 次 | 頁1-28 |
分類號 | 448.94 |
關鍵詞 | 認知科學; 腦電波; 線性鑑別分析; 主成份分析法; 經驗模態分解法; Cognitive science; Electroencephalography; EEG; Linear discriminate analysis; LDA; Principal component analysis; PCA; Empirical mode decomposition; |
語 文 | 中文(Chinese) |
中文摘要 | 大腦進行運作的機制一向是科學與醫學的研究重點,本研究以四位受測者為對象,進行想像左右手動及想向幾何左右旋轉之腦電波辨識率分類與比較。首先,利用主成份分析法及線性鑑別分析法做特徵擷取,接著使用最近鄰居法則做資料分類,實驗結果顯示四位受測者的平均辨識率可達85.75%。同時,使用非穩態訊號的經驗模態分解法(EMD)進行想像幾何左右旋轉之濾波,以線性鑑別分析法(LDA)與最近鄰居法(NNR)辨識想像順時針旋轉與逆時針旋轉的腦電波,實驗結果以FZ電極與FCZ電極的組合辨識率最高,四位受測者平均辨識率可達80.7%。對於大腦人機介面系統應用中,本研究有效提升想像左右手動的辨識率達10.07%,對於想像幾何左右旋轉亦首次建立可達80.7%的辨識率。 |
英文摘要 | The mechanism of brain function has been focus of intense scientific and medical research in recent years. Four subjects in this study were targeted and try to imagine the hand movements and the geometric rotation, and the corresponding electroencephalography (EEG) signals collected from the imagining activities are classified and compared. We use principal component analysis and linear discriminate analysis method to extract the EEG features, and then use the nearest neighbor rule to perform data average classification rate is up to 85.75%. For the imagining rotation, we use the empirical mode decomposition (EMD) of non-steady-state signals to filter the EEG, and then use linear discriminate analysis (LDA) and the nearest neighbor method (NNR) to classify the imagining clockwise rotation and counterclockwise rotation. The experimental results show that FZ electrode and electrode FCZ combination can be obtained a better classification rate, the average recognition rate is up to 80.7%. For human-machine interface system applications, the study can effectively upgrade the classification rate of the imagining hand movements about 10.07% rates. |
本系統中英文摘要資訊取自各篇刊載內容。