Prev: Date settings in Matlab and Access
Next: www.voguesneakers.com Cheap Jordan Shoes,Cheap Nike Shoes
From: Ahmad Alkandri on 21 May 2010 07:06
hi all

any ideas of how to measure half width and half height of a 2D correlation peak, where the peak is not too sharp and also its got a wide width ?

Ahmad
From: dpb on 21 May 2010 09:04
Ahmad Alkandri wrote:
> hi all
> any ideas of how to measure half width and half height of a 2D
> correlation peak, where the peak is not too sharp and also its got a
> wide width ?
> Ahmad

Typically one fits a distribution of some sort and uses the properties
of that distribution with the empirically-determined coefficients to do
the estimation.

W/O more to go on, don't know what more than such generalities one could
give. Somebody posted an example code not too long ago of using a
parabola w/ the normalization constants consistent w/ those of a
Gaussian; one can only surmise that was intended as an easily computed
approximation...

--
From: Ahmad Alkandri on 21 May 2010 17:27
i did find a function called fwhm(), which works with waveforms ... its not working well on the correlation output that i'm trying to mesure the Full width at half maximum


Ahmad
From: dpb on 21 May 2010 18:22
Ahmad Alkandri wrote:
> i did find a function called fwhm(), which works with waveforms ... its
> not working well on the correlation output that i'm trying to mesure the
> Full width at half maximum

All will be dependent on how well an assumed distribution fits the
observed data (or, theoretically, the other way 'round).

If you don't have any basis to select one over another, guess you could
start out w/ a period of distribution fitting/selection if the point is
to develop some method for future measurements of similar data.

--
 | 
Pages: 1
Prev: Date settings in Matlab and Access
Next: www.voguesneakers.com Cheap Jordan Shoes,Cheap Nike Shoes