dichotomic_grouping.m

matlab的工具箱

function [part,B,G] = dichotomic_grouping(pos,options)

% dichotomic_grouping - regroup points into 2^s (s is unknown) bins of nearly equal size. 
%
%   [part,B,G] = dichotomic_grouping_1d(pos,options);
%
%   'part' is a cell array containing the membership of each group.
%   'B' is a vector containing the membership of each point.
%   'G' is the number of points in each bin
%
%   'options' is a structure that can contains : 
%       - options.part_type: the way the dyadic split is performed, can be
%       either:
%           '1axis','2axis','3axis',etc : use the first 1,2,3,etc axis 
%              to do alternatively the split. Typically, if you have
%              3D data and you want approximatively isotropic grouping, 
%              use '3axis'.
%           'kmeans' : use a modified k-means method to do the grouping.
%       - options.ptmax: maximum number of point in each final cell.
%       - options.depthmax: depth of subdivision.
%       - options.part: an initial partition to refine.
%       - options.nb_iter : nbr of iterations for the K-means like clustering.
%
%   Copyright (c) 2004 Gabriel Peyr