Dear Paul,
the help message of DCT says:
...
N-1
X[k] = w(k) sum x[n] cos (pi (2n-1) k / 2N ), k = 0, ..., N-1
n=0
...
I think the term (2n-1) should be (2n+1). May I suggest to change the help
message.
(It's the same in IDCT)
Reason:
Trying to understand this formula as exactly as possible I wrote my own (very
simple, brute force, could be done much more sophisticated) program according
to that formula:
function x = myidct(X)
N = length(X);
NN = [1:N];
[k, n] = meshgrid(NN-1, NN-1);
wk = ones(size(NN)) * sqrt(2/N);
wk(1) = sqrt(1/N);
c = cos (pi * (2*n-1) .* k / (2*N));
s = X(k+1) .* c;
x = sum(repmat(wk', 1, N) .* s')';
It's results differ from DCT's.
Changing (2n-1) to (2n+1) my function produces the same results as DCT.
This is also conforming to the article at
http://en.wikipedia.org/wiki/Discrete_cosine_transform
Once again thank you for your great and fmaous work
with best regards
Torsten Finke
--
------------------------------------------------------------------------
Dr.-Ing. Torsten Finke Amtsgericht Essen HRB 11500
Ingenieurgemeinschaft IgH USt-Id.-Nr.: DE 174 626 722
Ges. für Ingenieurleistungen mbH Geschäftsführung:
Heinz-Bäcker-Str. 34 Dr.-Ing. S. Rotthäuser
D-45356 Essen Dr.-Ing. T. Finke
Tel.: +49 201 / 360-14-17 Dr.-Ing. W. Hagemeister
Fax.: +49 201 / 360-14-14 Tel.: +49 201 / 360-14-0
[email protected] http://www.igh-essen.com
------------------------------------------------------------------------
------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev
