Feature #120

Updated by herck almost 8 years ago

Under the assumption that the simulation model is able to capture all structural characteristics of the scattering image, one would expect that the difference image (simulation - real data) for the best fit contains relatively small low frequency components when transformed to the frequency domain.
One could implement this reasoning into the minimizer by creating a chi-squared function as follows:
# Compute the Fourier transform of the difference map (OK)
# Apply a filter to this image, thereby putting higher weight on the lower frequency components (only cutoff filter is implemented)
# Finally, calculate a norm of this filtered map that will serve as the minimizing function