查詢結果分析
來源資料
相關文獻
- 改良式對角化主要成份分析法應用於想像動作腦電波的辨識
- 『是非題』作答之腦電波辨識與『選擇題』作答之腦電波分析
- 以主成份分析法和線性鑑別分析法辨識想像左右手動
- 針灸抑制癲癇發作機轉之研究
- 類神經網路應用於腦病變源定位分析
- Valproate-Induced Encephalopathy
- 以腦電波探討具恢復力之環境對生心理影響之研究
- Brain Electrical Activity Mapping in Childhood Absence Epilepsy
- Design of on EEG Wavelet Prototype for Wavelet Analysis of EEG Focal Sharp Wave
- 大腦人機介面的實現--利用腦波P300訊號控制物體四個方向移動
頁籤選單縮合
題名 | 改良式對角化主要成份分析法應用於想像動作腦電波的辨識=Modified Diagonal Principal Component Analysis Applied to Motor Imagery EEG Classification |
---|---|
作者 | 張菀珍; 陳致仰; 蔡俊明; 葉榮木; Chang, Wan-jen; Chen, Chih-yang; Tsai, Chun-ming; Yeh, Zong-mu; |
期刊 | 明志學報 |
出版日期 | 20090600 |
卷期 | 41:1 2009.06[民98.06] |
頁次 | 頁19-32 |
分類號 | 448.94 |
語文 | chi |
關鍵詞 | 大腦人機介面; 對角化主要成份分析法; 腦電波; 時-頻-空域分析; Brain-computer interface; Diagonal principal component analysis; Electroencephalography; Time-frequency-spatial analysis; |
中文摘要 | 腦電波辨識是否成功的關鍵,在於特徵擷取與分類兩個議題。過去文獻將重點放在分類演算法的改良上,然而找出更簡單而重要的特徵,也可以獲得高辨識率。對角化主要成份分析法 (DiaPCA) 可以從腦電波高維度的時-頻-空域資料矩陣中找出主要的成份。被挑選出來的主要成份可構成一個較低維度的特徵矩陣,但仍保有兩種想像動作的腦電波之間主要的特徵差異。因此,藉由計算特徵矩陣間的歐式距離就可分類腦電波。本論文提出利用「改良式對角化主成份分析法」對腦電波擷取特徵並辨識,結果顯示,腦電波辨識的準確率大幅提升10.07%。 |
英文摘要 | Feature detecting and classifying algorithm are the two keys to Electroencephalogram (EEG) classifying. Past articles focused on the improvement of classifiers, but selecting simpler and more important features is an alternative way to get a high accuracy. Diagonal Principal Component Analysis (DiaPCA) can find out the pricipal components of the high-dimension Time-Frequency-Spatial matrix of EEG signals. The principal components construct a lower-dimension matrix, which carry the major differdnce between the two classes of EEG signals. Thus, we can easily the EEG sighals by calculating the Euclidean distance between feature matrices. A modified Diagonal Principal Component Analysis method is proposed in this paper to select feature vector and classify EEG. The result shows that the accuracy of EEG classifying is improved and up to 10.07%. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。