CodingGain

C = CodingGain(A);

  CodingGain(vars) returns the coding gain of a subband coding system of M 
  channels, where vars is an M-element vector of powers. 

  CodingGain(R) returns the coding gain of a subband coding system of M 
  channels, R is an MxMxL matrix expressing the polynomial covariance matrix
  of the subchannels. 

  The coding gain is the ratio between arithmetic and geometric mean of the 
  subband variances.

  Input parameter:
       vars or R  vector of subband powers / variances, or
                  polynomial covariance matrix

  Output parameter:
       C          coding gain

0001 function C = CodingGain(A);
0002 %C = CodingGain(A);
0003 %
0004 %  CodingGain(vars) returns the coding gain of a subband coding system of M
0005 %  channels, where vars is an M-element vector of powers.
0006 %
0007 %  CodingGain(R) returns the coding gain of a subband coding system of M
0008 %  channels, R is an MxMxL matrix expressing the polynomial covariance matrix
0009 %  of the subchannels.
0010 %
0011 %  The coding gain is the ratio between arithmetic and geometric mean of the
0012 %  subband variances.
0013 %
0014 %  Input parameter:
0015 %       vars or R  vector of subband powers / variances, or
0016 %                  polynomial covariance matrix
0017 %
0018 %  Output parameter:
0019 %       C          coding gain
0020 
0021 % S. Weiss, 25/8/2014
0022 
0023 [M,N,L] = size(A);
0024 if (L==1) && ((M==1)||(N==1)),
0025    J = max(M,N);
0026    C = (sum(A)/J) / ((prod(A)).^(1/J));
0027 elseif M==N,
0028    L2 = (L+1)/2;
0029    A = diag(A(:,:,L2));
0030    C = (sum(A)/M) / ((prod(A)).^(1/M));
0031 else
0032    error('input to CodingGain() does not satisfy the required dimensions');
0033    C=-1;
0034 end;
0035