function [qdata, qtable] = quantize(data, Ns, r, epsilon, dist2);
% quantizing data (in descriptor space) 
% see "Creating effivient codebooks for visual recognition" F. Jurie and B. Triggs
%
% I assume the data is a martix of (d)x(N) where d is the dimension of desc
%
% the parameters are:
% Ns - number of samples per mean_shift iteration
% r - cell radius  
% epsilon - epsilon accuracy of mean shift 
% A function pointer
% dist - distance measure between two descriptors - __RETURNS dist^2___
% 
% output:
% qdata - the quantized data
% qtable - table of quantized values 
%
%
%
% Copyright (c) Bagon Shai
% Department of Computer Science and Applied Mathmatics
% Wiezmann Institute of Science
% http://www.wisdom.weizmann.ac.il/
% 
% Permission is hereby granted, free of charge, to any person obtaining a copy
% of this software and associated documentation files (the "Software"), to deal
% in the Software without restriction, subject to the following conditions:
% 
% The above copyright notice and this permission notice shall be included in 
% all copies or substantial portions of the Software.
%
% The Software is provided "as is", without warranty of any kind.
%
% Sep. 2006
%


% fix the mean shift kernel width to be able to distinguish between two
% nieghbouring instances.
h = (r/2.0125).^2; % see gauss document...

% magic numbers
NUM_OF_BYTES = 2;
Q_SIZE = 2^(8*NUM_OF_BYTES);
convert = @uint16; % match the convert function to NUM_OF_BYETS

[d N] = size(data);

% init: population is all the possible indices
qdata = convert( zeros(N,1) );

% allocate NUM_OF_BYTES bytes for quant
qtable = zeros(d,Q_SIZE);
ii = convert( 1); % count into qtable

r2 = r.^2; % dist function returns dist^2

disp(sprintf('i=%5d',ii));

population = 1:N;
lp = length(population);

% while there's any zero value in qdata
while ( ( ii < Q_SIZE-1 ) & ( lp > 0 ) ) 

    % sample N data instances from non-clasified points
    samples = randsample(population, min(Ns, lp) );

    % find the mode using mean shift - note that inpt to mean sift is
    % (d)x(N) 
    mode = mean_shift(data(:,samples), h, epsilon, dist2);

    
    % remove all matching points from data (only where qdata = 0, i.e. in population)
    matches = find( dist2(data(:,population), repmat(mode,1,lp)) <= r2 );
    lm = length(matches);
    if ( lm == 0 )
        disp('no matches');
        continue
    end

    % insert this mode to qtable
    qtable(:,ii)=mode;

    qdata( population(matches) )  = ii;
    population(matches) = [];

    lp = lp - lm;
    ii = ii + 1;
    disp(sprintf('\b\b\b\b\b\b\b\bi=%5d',ii));
end
% resize qtable 
qtable = qtable(:,1:ii-1);