基於小波轉換及主成分分析之腦電訊號分類與移動方向辨識之應用

EEG Signals Classification Based on DWT and PCA for Direction Detection

林克勳(Ke-Syun Lin)

指導教授 ： 吳仁銘

國立清華大學/電機資訊學院/電機工程學系/碩士(2014年)

摘要

近年來由於穿戴式裝置的發展，腦電圖(Electroencephalography，EEG)以及大腦人機介面(Brain–computer interface，BCI)的相關應用成為熱門研究主題。大腦人機介面是一門嶄新的技術用於將受試者的意識轉換為對應指令，依此控制電子裝置、電子義肢及軟體應用。腦機介面使受試者僅需使用腦電訊號而不需透過肌肉與神經的傳遞方式與外界溝通，因此常用於老年人以及癱瘓患者。
為了完成以上應用層面的複雜工作，我們需要一套具相當可靠度的指令辨識方法。此類方法一般包含三部份：資料前置處理、特徵值擷取、資料分類模型。這篇論文提出一個腦電波特徵辨識方法，結合離散小波轉換 (Digital wavelet transform，DWT)、主成分分析 (Principal component analysis，PCA)、最短距離分類器 (Minimum distance classifier)，這套方法用於辨識受試者想像方塊的四個移動方式(提起、中立、推遠、拉近)。本研究以Emotiv頭戴式腦波儀擷取受試者腦電圖原始資料作為資料庫，於不同日期蒐集同一受試者腦電波資料以期獲得平均辨識率。實驗部分以多種方法的辨識率比較證實所提出的方法能獲得較高的辨識效能。
經由適當的訓練次數，此分類方法對於方塊移動方向的辨識可達到92.5%的準確度。本研究進一步以文獻探討方式，比較所提出方法與其他大腦人機介面研究的辨識率：同樣使用Emotiv作為腦波儀器，本方法能獲得較高辨識率；使用其他腦波儀的研究方面，本方法亦能達到接近的辨識率。

關鍵字

腦電圖 ； 大腦人機介面 ； 離散小波轉換 ； 主成分分析

並列摘要

In recent years, Electroencephalography (EEG) and Brain–computer interfaces (BCIs) have been widely discussed and become a popular research topic. The BCIs are innovative interfaces for translating the user’s intent into commands to control devices or software applications with recognized symbols. Via using BCI systems, it is capable to recognize the brain signals without the dependence of the normal output pathways of the peripheral nerves or muscles.
To accomplish such complex tasks, a pattern recognition algorithm with high reliability is especially needed. Such algorithm basically consists of three parts: data pre-processing, dimensionality reduction, and classification model. In this thesis we proposed a pattern recognition method that used to identify intention of movement from the users. The method is composed of digital wavelet transform (DWT), principal component analysis (PCA), and minimum distance classifier.
The proposed method is designed to recognize the subject’s intention of moving a cube. The subject is asked to perform 4 types of mental tasks related to movement, which are lift, neutral, push, and pull. With the proposed pattern recognition method and proper training, the EEG raw data recorded by an EEG headset can be precisely classified into corresponding command of movement with high recognition rate of 92.5%.

並列關鍵字

Electroencephalography (EEG) ； Brain–computer interfaces (BCIs) ； digital wavelet transform (DWT) ； principal component analysis (PCA)

參考文獻

[1] L. F. Haas, “Hans Berger (1873-1941), Richard Caton (1842-1926), and electroencephalography,” Journal of Neurology Neurosurgery and Psychiatry, 741:9, 2003.

[2] A. Lenhardt, M. Kaper, and H. J. Ritter, “An Adaptive P300-Based Online Brain–Computer Interface,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 16, no. 2, pp. 121–130, Apr. 2008.

[3] J. N. Mak, J. R. Wolpaw, “Clinical Applications of Brain-Computer Interfaces: Current State and Future Prospects,” IEEE Reviews in Biomedical Engineering, vol. 2, pp. 187–199, 2009.

[4] B. Rebsamen, C. Guan, H. Zhang, C. Wang, C. Teo, M. H. Ang, and E. Burdet, “A Brain Controlled Wheelchair to Navigate in Familiar Environments,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 18, no. 6, pp. 590–598, Dec. 2010.

[5] R. Beisteiner, P. Höllinger, G. Lindinger, W. Lang, and A. Berthoz,“Mental representations of movements. Brain potentials associated with imagination of hand movements,” Electroencephalography and Clinical Neurophysiology, vol. 96, pp. 183–192, 1995.