Prev: alternative 'linspace'
Next: nonlinear curve fitting choices
From: Kang Soo Lee on 2 Jun 2010 14:39
Hello.

I'm new at Matlab.
These days, I'm trying to fit several peaks (~30 peaks) into several Gaussians to the whole array in 2D image which has dimension 313*265.

However, so far, my code can fit only one peak.
Can any one help me to find out the method?

With best regard.
From: ImageAnalyst on 2 Jun 2010 18:36
On Jun 2, 2:39 pm, "Kang Soo Lee" <ks-...(a)kaist.ac.kr> wrote:
> Hello.
>
> I'm new at Matlab.
> These days, I'm trying to fit several peaks (~30 peaks) into several Gaussians to the whole array in 2D image which has dimension 313*265.
>
> However, so far, my code can fit only one peak.
> Can any one help me to find out the method?
>
> With best regard.

---------------------------------------------------------------
Maybe you can try the "CLEAN" algorithm - it's used in astronomy to
find stars in photos of the night sky.
I think it basically works like this:
1. find the brightest peak.
2. if no peak is big or bright enough, exit loop
3. take neighborhood around the peak and model it to a Gaussian
4. write modeled Gaussian to an output image.
5. subtract modeled Gaussian from input image.
6. repeat steps 1-5

Basically you're adding perfect Gaussians to your output image and
removing them from your input image (hence the name "clean").
Good luck,
ImageAnalyst
 | 
Pages: 1
Prev: alternative 'linspace'
Next: nonlinear curve fitting choices