nlinfit for a function of 2 variables

Hey,

I want to get a Gaussian git for my data using Matlab. However, I'm expecting a 2D Gaussian - Z(x,y)=A*exp(-(x^2+y^2)/2σ^2) Is there a way of using nlinfit for this?

I saw another question here about nlinfit for a 2 variables function, however in that function they weren't coordinates. I need all the combinations of X and Y in order to make the fit.

loc=[XC YC];
  gaussian = @(param,x) param(1)*exp( -((x(:,1)-param(2)).^2+(x(:,2)-param(3)).^2+)./(2*param(4).^2)  ) + param(5);
  params=zeros(1,5);
  params(1)=A*maxVal;
  params(2)=maxXpos;
  params(3)=maxYpos;
  params(4)=estW;
  params(5)=nPixInf(1,1);
  paramsX = nlinfit( loc , nPixInf , gaussian , params );

XC and YC are just 2 one dimensional arrays that goes from 1 to the size of that dimension. nPixInf is a 2 dimensional matrix which suppose to look like a Gaussian.

Can I use nlinfit for this?

Thanks!

Answer 1

[Star Strider]

It looks to me that you have coded your ‘gaussian’ function correctly. (I’ve used this approach successfully with nlinfit.)

Are you having problems with it?