A second check of the quality of a cleaned image: CHI square test
Next to the comparision between the component map
and the residul map, the quality of the cleaned images can also be tested
with a CHI square test.
The Standard Example: two gaussian sources
IDL> o = hsi_image()
IDL> cleaned_map = o -> getdata(image_alg='clean',clean_no_chi=0)
by setting the keyword 'NO_CHI' to zero, a window is displayed showing
the modulation pattern (blue = observed, white = clean components):
some more information about the displayed maps:
- two averaged values with different weighting (NATURAL and UNIFORM) of the subcollimators are displayed.
- the observed counts are the total counts observed, e.g. the total of the blue curves
- the clean counts are the total counts in the component map, e.g. the total of the with curves
- the evolution of the CHI2 values as a function of the number of iteration is displayed
in the top right corner.
Try to enhance the cleaning by playing around with the following parameters:
- try to use clean windows, e.g. set the keyword MARK_BOX.
- use a smaller gain to clean, e.g. set the keyword FRAC to 0.1 or even smaller.
QUESTIONS/COMMENTS/ETC. to Säm Krucker, email@example.com, 510 643 3101