BackgroundNeutralization question

[RBA]

Does the BN process use the values of every pixel above the lower limit and below the upper limit to do its calculations? If not, what else determines what is to be considered background and what isn't?

[RBA]

Nobody knows? 

[georg.viehoever]

Have a look at the documentation 

[RBA]

I have read the docs, Georg. In fact, I actually put together some documentation myself, remember?
http://blog.deepskycolors.com/PixInsight/ColorCalibration.html#BackgroundNeutralization

But if that's the case, then the default value of 0.1 for the upper limit seems rather aggressive, doesn't it?
At least, in pretty much all my images, after calibration/etc, most of the data - basically everything except the core of stars - is always below the 0.1 threshold, so applying BN with the default values would be technically incorrect. That's why I was asking the question.

And yes, I understand a default is just that, so maybe I'm just thinking too much here...  8)

[Carlos Milovic]

Hi Rogelio
I have not seen the code of BN, but those values certainly look like thresholds of the data, to set valid samples, and nothing else.
Also, there seems to be no statistical rejection, to avoid stars for example. The process is just confident that you selected a good reference, where the background dominates the statistics (the peak of the histogram is a good approximation to the background mean value, and so is the median of the reference).

You may decrease that value to avoid "contamination" of the sample, and let the median to be closer to the peak, but I would say that for background neutralization purposes, it is not critical.

[georg.viehoever]

I have read the docs, Georg. In fact, I actually put together some documentation myself, remember?
http://blog.deepskycolors.com/PixInsight/ColorCalibration.html#BackgroundNeutralization

...

Hi RBA,
My intention was to remind the PTeam that (official) documentation is very important. It should be possible to answer such questions without looking at the source code.
Georg

[RBA]

You may decrease that value to avoid "contamination" of the sample, and let the median to be closer to the peak, but I would say that for background neutralization purposes, it is not critical.

I agree with that. I guess I' m thinking in a more theoretical fashion.
The general recommendation so far seems to have been to mainly "select a reference that is well representative of the background", but I have never seen a recommendation of for example, actually looking at the histogram to better determine - even approximately - where the background ends and the signal begins, .

As I see it, usually the background seems to sit at a much lower range than "everything above 0 and below 0.1". Not a problem indeed if a good reference is selected, when a good reference can be found.

Thanks!

[martin_magnan]

Well... sometime  simple trivial question can lead to enlightenment. that just happen to me with this thread. I had mixed success with the neutralize background process. I was using default values Now everything is clear.

Thanks !!!
MArtin