Mercurial > hg > ltpda

comparison m-toolbox/classes/+utils/@math/spflat.m @ 0:f0afece42f48

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Import.
author Daniele Nicolodi <nicolodi@science.unitn.it>
date Wed, 23 Nov 2011 19:22:13 +0100
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:f0afece42f48
1 % spflat measures the flatness of a given spectrum
2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
3 %
4 % DESCRIPTION: spflat measures the flatness of a given spectrum using the
5 % spectral flatness measure method reported in [1]. It outputs the spectral
6 % flatness coefficient:
7 %
8 % 0 <= sw <= 1
9 %
10 % If the spectrum is peaky then sw is near 0. The more the spectrum is flat
11 % the more sw in near to 1. The sw coefficient is practically calculated by
12 % the ratio between geometric mean and arithmetic mean of the spectrum.
13 %
14 % CALL:
15 %
16 % sw = spflat(S)
17 %
18 % INPUTS:
19 %
20 % - S sample power spectrum. More than one sample spectrum can be
21 % input if they are combined in a single nxm matrix. The algorithm
22 % calculates sw for each spectrum.
23 %
24 % OUTPUT:
25 %
26 % - sw spectral flatness coefficient. If more than one spectrum
27 % are input sw is a row vector.
28 %
29 % REFERENCES:
30 %
31 % [1] S. M. Kay, Modern spectral estimation:Theory and Application,
32 % Prentice Hall, Englewood Cliffs (1988) ISBN-10: 0130151599. Pages
33 % ??.
34 %
35 %
36 %
37 % VERSION: $Id: spflat.m,v 1.2 2009/01/29 08:59:26 luigi Exp $
38 %
39 %
40 % HISTORY: 28-01-2009 L Ferraioli
41 % Creation
42 %
43 %
44 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45 function sw = spflat(S)
46
47 % willing to work with columns
48 [a,b] = size(S);
49 if a>1 && b>1
50 warning('Matlab:MultipleSpectra','A matrix of data was input; Spectral Flatness Coefficient will be calculated for each column')
51 else
52 if a<b
53 S = S.';
54 end
55 end
56 Ns = size(S,1);
57 gmean = exp(sum(log(S))./Ns);
58 sw = gmean./mean(S);
59
60 end