entropiv

PURPOSE ^

'ENTROPY' calculates the entropy of an isolated matrix referring it

SYNOPSIS ^

function [e,ta]=entropia(a)

DESCRIPTION ^

 'ENTROPY' calculates the entropy of an isolated matrix referring it
 to its own statistics, i.e. account all the different values
 (Xi) and the number of occurrences (ni) and calculate the probability
 of each xi as Pi = ni / N.
 At the end makes the corresponding summation:
             _
            |
 Entropy =  | P (x) log2 (P (x)) dx
            |
           -

 USE: [e] = entropia(a);

CROSS-REFERENCE INFORMATION ^

This function calls: This function is called by:

SOURCE CODE ^

0001 % 'ENTROPY' calculates the entropy of an isolated matrix referring it
0002 % to its own statistics, i.e. account all the different values
0003 % (Xi) and the number of occurrences (ni) and calculate the probability
0004 % of each xi as Pi = ni / N.
0005 % At the end makes the corresponding summation:
0006 %             _
0007 %            |
0008 % Entropy =  | P (x) log2 (P (x)) dx
0009 %            |
0010 %           -
0011 %
0012 % USE: [e] = entropia(a);
0013 
0014 function [e,ta]=entropia(a)
0015 
0016 n=size(a);
0017 k=1;
0018 
0019 for i=1:n(1)
0020     for j=1:n(2)
0021        dato=a(i,j);
0022        if (i==1)&(j==1)
0023           ta=[1 dato 1];
0024        else
0025           valores=ta(:,2);
0026           condicion=sum(((valores-dato)==0));
0027           if condicion>0
0028              indice=find((valores-dato)==0);
0029              ta(indice,3)=ta(indice,3)+1;
0030 
0031           else
0032              k=k+1;
0033              ta(k,:)=[k dato 1];
0034 
0035           end
0036        end
0037    end
0038 end
0039 
0040 P=ta(:,3)/(n(1)*n(2));
0041 LP=-log(P)/log(2);
0042 
0043 e=sum(P.*LP);
0044

Generated on Fri 07-Mar-2014 13:29:20 by m2html © 2005