Questions about Jordi Gallego's Image Integration Presentation

[f11]

I've been going through Jordi's Image Integration presentation powerpoint slides in an effort to see if I can increase the SNR of my noisy images. In it he describes a process of repeating the integration of the registered frames using different rejection algorithms, looking for the one that provides the best SNR. He then takes the viewer through a fine tuning process to find the optimal rejection settings for the selected algorithm.

What I don't get is: in his first example of integrating 15 images, he shows the noise stats for integrating with Sigma, Winsorized Sigma and Linear Fit Clipping algorithms, in which LFC had less Guassian noise from the outset, and a better SNR than the other two choices.  Yet, three slides later, he chooses the WSC as his preferred rejection algorithm. Other than one dark pixel being visible in the other three that isn't in the WSC, on what basis did he select that algorithm?

What was the point of the SNR determination if the result it suggests is itself rejected? I understand his approach and his method, but I sure don't understand his conclusion regarding the optimum rejection algorithm.

Jordi is obviously a smart guy, and I'm not questioning his science... so I must have missed something that would explain his decision... but I can't figure out what.  Can anyone shed some light on this?

Rod

[Josh Lake]

Do you have a link?

[f11]

Um, sorry - I thought everyone was aware of the summary page of good PixInsight links that Drake Visual made available at

< http://www.drakevisual.com/pi-links-by-topic > ...

A guy could spend a week or two just reading and watching the great info that's available on this single page.


Anyway - the specific powerpoint presentation that I'm referring to is here:

< http://www.astrosurf.com/jordigallego/articles/Image_integration_JGallego.ppt >

Rod

[Jordi Gallego]

Dear Rod, I am sorry but I missed this post (I have not been very active later on ).

This presentation was made for an audience of about 60 people and there was a lot of verbal discussion and questions at the end. In my website I also included a text file with most of my explanations to the audience:

http://www.astrosurf.com/jordigallego/articles/Integracion_imagenes_JGallego_notas.pdf

These are 8 pages of text, only in Spanish and I apologize I have no time now to properly translate to English. However I guess you still can get useful information from this document trough an automatic Goggle translation.

In any case, to provide you an answer to your question, I give you a short translation of my explanation of dispositive 35:

“We take a look to the 4 zooms at 300%.
On the Average image we can see almost everything: we have the trace of the plane, we see a second and weaker plane trace, almost horizontal, hot pixels here and there and also cold pixels.
Looking to the other 3 images we see that traces of airplanes and hot pixels are gone in all 3. But with regards to cold pixels, Sigma Clipped image still retains many (in fact all of them, since we have seen previously that there was no cut at all), Linear Fit image is much better, but shows two black dots that are quite visible. Winsorized image, has on the contrary very soft sky background and presents no cold pixel.
Therefore (and having done this check on other key points of the image) I select Winsorized Sigma Clipping algorithm.”
 
I apologize for my bad translation and hope that can help you a bit

Best regards

[f11]

Thanks, Jordi - that helped a lot.  I'll take a look at the links you provided as well.  Greatly Appreciated!!

Rod