Author Topic: Registering images with PixInsight  (Read 4217 times)

Offline cweinhe

  • Newcomer
  • Posts: 2
    • View Profile
Registering images with PixInsight
« on: 2006 August 06 21:31:16 »
All I can say is PixInsight is amazing.  As a scientist, I appreciate the quality of the software and it is very intuitive to me.  That said, I could always use more features...  :wink:

I've successfully stacked some images and processed them using the Histogram feature, curves, and the DBE (flat fielder) with some very good results. I'll post them later after I take another run at it.

I would like to see a feature added to PixInsight that would allow me to shift an image in X and Y before adding it to another image.  With rotation and this feature I could stack registered images knowing the dX, dy and angle of rotation of each frame.  I have a program that registers images, but it sucks at combining them.

I was going to buy photoshop so I could process my astrophotography photos, but I no longer feel that I need to with PixInsight.

Thanks for all your hard work.


Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 7097
    • View Profile
Registering images with PixInsight
« Reply #1 on: 2006 August 08 10:46:29 »
Hi Corey,

Thank you for the nice words  :D

In my answer to your question on the General Discussion forum, I've just exposed our plans regarding image registration. We'll have both automatic and a semi-manual (or is it semi-automatic?  :lol: ) registration processes.

Have you seen the ChannelMatch process in the PixInsight Standard beta? If you are willing to tolerate a bit of trick, it can be used to do what you want very easily. This process is actually intended to align individual RGB channels in a color image. However, nothing stops you to specify the same displacements to the three RGB channels of an image, which in fact applies a translation to the whole image.

The ChannelMatch process pertains to the Geometry category; you'll find it on the Process Explorer window, or by Process > Geometry > ChannelMatch, from the main menu.

Having said that, it's pretty obvious that a specific process for image translation should be included in the Geometry category. Thanks for showing this to us.

One last thing about ChannelMatch. This process allows you to perform sub-pixel translations, but if you apply integer offsets (X-Offset and Y-Offset parameters), the process does not interpolate pixel values since the translation work can be done by simply copying pixels. This is important because interpolation must be avoided, as far as possible, especially at initial processing stages.

To integrate your images (a.k.a. stack or combine), you can use PixelMath. The new PixelMath, also available in the PixInsight Standard beta, allows mathematical expressions in common algebraic notation. For example, to average three images, you can apply this expression:


with the Rescale Result option enabled. If you want to apply different weights to individual images, you could write something like:

Image01 + 0.95*Image02 + 0.87*Image03

Beware of the new PixelMath: it is highly addictive. If you find yourself playing with a bunch of images during hours and hours, trying out innumerable expressions, don't say I didn't warned you :lol:

Juan Conejero
PixInsight Development Team