English Deutsch Français Italiano Español Português 繁體中文 Bahasa Indonesia Tiếng Việt ภาษาไทย
All categories

Hi. I'm trying to deconvolve two functions (form: ax^b exp(-x/c) ) with each other. I'm using the discrete fourier method and MATLAB. Sometimes I get a peak, and with other curves I get a peak followed by a small valley. Where does this valley come from? How can I reliably deconvolve these curves? thanks.
please see:
http://s153.photobucket.com/albums/s235/s1020099/

2007-01-24 22:30:51 · 1 answers · asked by Peter R 2 in Science & Mathematics Physics

1 answers

It would appear that you are curve fitting raw data and then attempting to deconvolve the analytic curve. You should do the deconvolution on the raw data to enhance detail and then curve fit after ward. For deconvolution to work well, you need very good signal-to-noise in the original data because the deconvolution blows up the noise (which you don't have). The side lobe is a symptom that one of the parameters in the assumed function is a bit aggressive. You can back off on the deconvolution enhancement factor a bit and the lobe will go away. I like to see a small lobe (<5-10% of the peak max) because it is a sign that the resolution enhancement has gone as far as it can. Further enhancement produces garbage.

2007-01-25 16:48:39 · answer #1 · answered by d/dx+d/dy+d/dz 6 · 0 0

fedest.com, questions and answers