Improved EEG Feature Extraction for BCI Based on RCSP

Brain Computer Interfaces (BCI) are provide a communication channel between human and computer. Many of this systems have been built on the basis of mental activity. One of the most popular feature extraction algorithms for BCI is the Common Spatial Patterns (CSP) algorithm. this paper present a method for imagery movement based on Regularized Common Spatial Pattern (RCSP) for feature extraction. We also propose 3 other methods based on RCSP and finally compared results. We have implemented proposed algorithm on EEG data of 9 subject, from BCI competition data sets. Result showed taht CSP with Tikhonov Redularization (TRCSP) and Weighted Tikhonov Regularization(TRCSP) are the best algorithms for this aim

کلیدواژه ها:

Brain Computer Interface ، Common Spatial Pattern ، Regularized Common Spatial Pattern ، Tikhonov Regularization Common Spatial Pattern ، Weighted Tikhonov Regularization Common Spatial Pattern