LabVIEW
Guaasian Peak fit results differ from other fiitting software results
Solved!
Ahh thanks. I did not notice the default value for the offset bounds and thought it's also +-inf. 😉
Yes, now the results agree for the fit that include an offset. Curiously, if I remove the offset from the nonlinear model (only 3 parameters) and don't wire the bounds for the stock gauss fit (to fix the offset at zero), the nonlinear filt seem significantly better (residue=3493.1) than the gauss fit (residue=5820.45). Shouldn't they agree here too??
@Aojha wrote:
I have attached a n image.
That seems to be a very different problem. Your ~70 bright spots seem to be 2D gaussians with two different widths and they seem to be arranged on a very regular grid.
Maybe you could do a 2d FFT to get the main spacing in each dimension, then slice the picture into regular little squares, each containing one 2D gaussian and fit each seperately for widths and intensity.
