5 replies to this topic

[KWhitt]

How can I center full-disc solar images, from a python script?

 

 

What I'm trying to do (big picture):  Process a year's worth of (mostly) daily white light images of the sun, to use in creating a timelapse animation.

 

I'd love if there was a command line interface to PIPP, because the center and crop functionality it offers is otherwise exactly what I want. (I know PIPP can process multiple images at once, but I want to script the whole process so adding a new day's image to the animation requires as few manual steps as possible.)

 

I've tried centering based on average brightness, but the results aren't good enough.

 

I've dabbled with detecting circles with the opencv/cv2 Hough Transform, but haven't managed to tune the parameters to get the results I want yet.  Before I spend more time on that path -- is there something else I should try?

 

Thanks,

Kristie

[gmiller123456]

Using OpenCV, I'd start by doing a threashold to separate the background/foreground into black and white, then play with blob detection to find the center of the biggest blob.

[KWhitt]

Ooohhh, threshold sounds promising.  Where my efforts have been inadequate so far seems to be from a slight gradient outside the disc, at least sometimes due to haze--a threshold would definitely help there.  I don't know much about blob detection, but I didn't know about Hough circles before today, either, so that part doesn't worry me.

 

Thanks!

[RobertJ]

This might have what you're looking for - either use the tool entirely, or be... inspired... by the code

 

https://github.com/GregAlt/SolarFinish (fantastic tool!)

 

The author posts here

 

-- Robert

[spicerack0]

go on xai and just ask grok to write you a code to center all your images around the center of contrast.

 

link to the response and sample code here- https://grok.com/sha...1c-fe7350c2f56a

Edited by spicerack0, 21 March 2025 - 03:11 PM.

[KWhitt]

This might have what you're looking for - either use the tool entirely, or be... inspired... by the code

 

https://github.com/GregAlt/SolarFinish (fantastic tool!)

That tool looks great, thanks!.  I'm away from home (and my data), but I'll have to give it a try on my h alpha stuff when I get back.

 

As for my centering, I gave the Hough circles another try and got parameters that work on my set of test images.  Full run on all my data when I get home again.

 

My next wish is that TlitingSun had a command line option, so I could figure out how much to rotate each SeeStar image so that solar north is up.  I can get the value of P from the sunpy package, but I'm puzzling over the N-S meridian tilt value.  All my searches keep giving answers to questions I'm not asking (ex: how to calculate the declination of the sun, or how to calculate the hour angle of the N-S meridian), and I don't know enough to build this up from scratch.  It looks like the SolarFinish tool does rotation by matching to a known image, which is clever!--but I'm determined to do it by math (I also don't know that would work with white light images).