Author Topic: Help needed: Registration of 2 Images with different scales/rotation  (Read 3745 times)

Offline Christoph Puetz

  • PixInsight Addict
  • ***
  • Posts: 150
  • Peterberg Observatory (Germany, Saarland)
    • View Profile
    • Fotos
Dear PI users, I now own PI for more than a year (maybe 2?) and I had some wonderful experiences
with it - great tool of software.
Now I have a problem where I get "stuck":

With different users, cameras and optics, we had multiple exposures of the region of the seagull-nebula.
As you may see in the attachment, we have a mosaic of two frames, each taken with a SKYWATCHER ED80 wih f=510mm
and a detail of IC2177, taken with a 6" Starfire refractor at f=1100mm.

Don't care about the colours of the mosaic, it is still in progress.
My problem is:
Is it possible to rotate/scale/register the detail-view of IC2177 into the "great" mosaic of the seagull nebula ??
I tried FFTRegister and a lot of other functions, but I did not succeed. So any help is appreciated.  ;)

You may download the 2 full-size-FITS-files with the following URLs:

File seagull_mj_cp: https://www.hidrive.strato.com/lnk/8WMEXCoK
File k-gum_v1.fit: https://www.hidrive.strato.com/lnk/dDsE3bWF
(currently the links are valid for 40 days)

Thanks in advance for your help.
Christoph
Kind regards,
      Christoph
---
ATIK 383L+, Canon EOS 450d, modified,
Canon EOS 500d, 
20" Planewave CDK, 6" APO Starfire Refractor,
Celestron 8", Skywatcher ED80,
Peterberg Observatory (www.sternwarte-peterberg.de)
PixInsight, PHD-Guiding
private URL: www.ccdsky.eu

Offline georg.viehoever

  • PTeam Member
  • PixInsight Jedi Master
  • ******
  • Posts: 2129
    • View Profile
Good evening Christoph,
 
It worked for me by with the following setting (see screenshot):

- set mosaic as reference image
- define preview on approximate target region (to give a hint to StarAlignment)
- deactivate "Use brightness relations" (because images are not linear) and "Use scale difference" (because images have different scale)
- apply to detail.

From my experience, StarAlignment works better on linear than on processed images. So you may want to do the alignment on linear images.

Georg
« Last Edit: 2012 March 26 08:01:22 by georg.viehoever »
Georg (6 inch Newton, unmodified Canon EOS40D+80D, unguided EQ5 mount)

Offline Andres.Pozo

  • PTeam Member
  • PixInsight Padawan
  • ****
  • Posts: 895
    • View Profile
I can confirm the hints of Georg. I have also been able to align the detail image using the mosaic as a reference after defining a preview around the area of the detailed image.
Code: [Select]
StarAlignment: Processing view: k_gum_v1
seagull_mj_cp: Restricting star matching to 1 preview(s).
seagull_mj_cp:
2102 stars found.
k_gum_v1:
Structure map: 100%
Detecting stars: 100%
3335 stars found.
Matching stars ...
* Reference image: Limiting to 2000 brightest stars.
* Target image: Limiting to 2000 brightest stars.
518 putative star pair matches.
Performing RANSAC ...
* Previous attempt failed - this is try #2
useScaleDifferences=false
* Reference image: Limiting to 2000 brightest stars.
* Target image: Limiting to 2000 brightest stars.
27 putative star pair matches.
Performing RANSAC ...
22 star pair matches in 190 RANSAC iterations.
Summary of model properties:
Inliers     : 0.815
Overlapping : 0.784
Regularity  : 0.739
Quality     : 0.793
Root mean square error:
?RMS  :  0.330 px
Average RMS error deviation:
?RMS  :  0.172 px
Peak errors:
?xmax :  0.462 px
?ymax :  0.804 px
Transformation matrix:
     +0.7131     -0.2852  -1048.0163
     +0.2784     +0.7100   -844.0357
     -0.0000     -0.0000     +1.0000
scale    : 1.324
rotation :    +21.85°
dx       :  -1048.02 px
dy       :   -844.04 px
Generating registered image: k_gum_v1_registered
Homographic Projection / Lanczos-3 interpolation, c=0.30: 100%
Registration successful.
19.02 s

ruediger

  • Guest
Good evening Ruediger,
Good morning Georg,
guess you meant Christoph :-)

My earlier problems with DynamicAlignment and mosaic building are solved, my first mosaic is finished and I'm very happy with the overall result. Thanks for the Gradient Domain Processes, they work great!

Rüdiger

Offline Christoph Puetz

  • PixInsight Addict
  • ***
  • Posts: 150
  • Peterberg Observatory (Germany, Saarland)
    • View Profile
    • Fotos
Good evening Georg, Rüdiger and Andres.
I tried your hints - and in fact IT WORKED !

Again there were some new things to learn about PI Star Alignment.
Thank you very much for your support !
 8)
Kind regards,
      Christoph
---
ATIK 383L+, Canon EOS 450d, modified,
Canon EOS 500d, 
20" Planewave CDK, 6" APO Starfire Refractor,
Celestron 8", Skywatcher ED80,
Peterberg Observatory (www.sternwarte-peterberg.de)
PixInsight, PHD-Guiding
private URL: www.ccdsky.eu