perform_local_pca.m

toolbox_dimreduc - a toolbox for dimension reduction methods This toolbox is an educational and rec

function [LocP,LocPsi,Segm] = perform_local_pca(X, options)

% perform_local_pca - perform local principal component
%
% [LocP,LocPsi,Segm] = perform_local_pca(X, options);
%
%   X is the data set of size (m,p), X(:,i) is the ith point 
%       leaving in dimension p.
%
%   LocP(:,:,i) is the matrix whose d columns are the basis
%       of the tangent plane in the ith cluster.
%   LocPsi(:,i) is m dimensional vector which is the center 
%       of the ith cluster.
%   Segm(i) tels to which cluster belongs the ith point of X. 
%
%   You can define
%       options.nbr_clusters : number of center points
%       options.nb_iter : number of iteration for the clustering.
%       options.nbr_neighbors : number of NN for tangent planes estimation
%       options.dim=d : dimensionality of tangent planes
%
%   Only a k-means clustering is used to perform segmentation.
%   
%   Copyright (c) 2006 Gabriel Peyr