mahalan.m

很好用的统计模式识别工具箱

function dist = mahalan(X,mi,sigma)
% MAHALAN Mahalanobis distance.
% dist=mahalan(X,mi,sigma)
%
% MAHALAN calculates Mahalanobis distance for given vectors
%   in the matrix X, where mi is the vector from which the distance
%   is calculated and sigma is the covariance matrix (bending of
%   the feature space).
%
% Input:
%   X [NxK]  matrix of K vectors for which the distance is computed,
%     where N is the dimension of the feature space.
%   mi [Nx1] mean vlaue. 
%   sigma [NxN] covariance matrix.
%
% Output:
%   dist [1xK] vector of distances for given vectors in the matrix X.
%

% Statistical Pattern Recognition Toolbox, Vojtech Franc, Vaclav Hlavac
% (c) Czech Technical University Prague, http://cmp.felk.cvut.cz
% Written Vojtech Franc (diploma thesis) 19.3.2000
% Modifications
% 27-Oct-2001, V.Franc
% 25. 6.00 V. Hlavac, comments into English.


% gets # of points
K=size(X,2);

% gets dim
DIM=size(X,1);

% makes mi a column vector
mi=mi(:);

% computes it for all the points in matrix X
if DIM == 1,
   dist=((X-repmat(mi,1,K))'*inv(sigma).*(X-repmat(mi,1,K))')';
else
   dist=sum( ((X-repmat(mi,1,K))'*inv(sigma).*(X-repmat(mi,1,K))')');
end