Hi Wally, This paper by Donoho is publicly available from Biometrika and has the example I was talking about:
http://biomet.oxfordjournals.org/content/81/3/425.abstract The algorithm I tried is the one he calls VisuShrink. It's described more rigorously in a paper in Information Theory (IEEE): http://ieeexplore.ieee.org/search/srchabstract.jsp?navigation=no&arnumber=382009 but you need a subscription to access the full text. I find these papers somewhat hard to follow (not my field) but the algorithm is pretty simple: wavelet transform, reduce the magnitude of the coefficients by a magic number related to the noise level, and inverse transform. Regards, Brian Date: Tue, 31 May 2011 15:30:22 +0200 > From: wally <wa...@voosen.eu> > To: help-gsl@gnu.org > Subject: Re: [Help-gsl] peakfind using GSL > Message-ID: <201105311530.22449.wa...@voosen.eu> > Content-Type: Text/Plain; charset="iso-8859-1" > > Hello Brian, > > thanks for this information. > may you please add a link to the example you used. > thx > wally > > _______________________________________________ Help-gsl mailing list Help-gsl@gnu.org https://lists.gnu.org/mailman/listinfo/help-gsl