CSP 共空间模式的 Matlab代码实现

Colbie

CSP 共空间模式的 Matlab代码实现

• function W = csp(X,Y)
• % Common spatial patterns for spatial filtering
• %
• % X: EEG data of class 1 (channel x sample point x trial)
• % Y: EEG data of class 2 (channel x sample point x trial)
• % W: the colunms of projection matrix W are the spatial filters
• % d: eigenvalues
• %
• % yuzhang 2013.12.16, ECUST, China
• %
• %% Compute covariance matrix for two classes
• Cx = zeros(size(X,1),size(X,1));
• Cy = zeros(size(Y,1),size(Y,1));
• for i = 1:size(X,3)
•     Cx = Cx + cov(X(:,:,i)');
• end
• for i = 1:size(Y,3)
•     Cy = Cy + cov(Y(:,:,i)');
• end
• Cx = Cx/size(X,3);     % covariance matrix averaged on all trials
• Cy = Cy/size(Y,3);
• C=Cx+Cy;
• %% Solve CSP+FC joint spatial filtering
• %[eigvec_C,eigval_C]=eig(C);
• %P=eigval_C^(-0.5)*eigvec_C';
• %S1=P*Cx*P';
• %[V,eigval_S1]=eig(S1);
• %[sort_val,idx]=sort(diag(eigval_S1),'descend');
• %final_eigvec=V(:,idx);
• %W=(final_eigvec'*P)';
• [W,D] = eig(Cx,Cx+Cy);
• [d,idx] = sort(diag(D),'descend');   %eigvectors sorted via eigvalues descend
• W = W(:,idx);
  6 a+ q) d: i3 b& X, g0 k

1 w+ P  C( B' ~# b0 l

BUFSFER

CSP 共空间模式的 Matlab代码实现。