Image alignment artifacts (i.e. ringings): how to reject them

[lucchett]

Hello guys,
I am having issues in getting the most from the Image Registration Module.

Lanczos produces ringing ( on my images) on small -medium stars that I can't tame with the clamping threshold.
the others method like bicubic spline give soft details.

I wonder if someone can give some advice about how to reject the artifacts I get with Lanczos.
I usually integrate the aligned frames with windsorized sigma clipping, but can't get rid of ringing and distorted stars (slightly rombic shaped).

Many thanks in advance,
Andrea

[mschuster]

Hi Andrea,

Have you tried one of the methods that does not ring as much followed immediately by a sharpening process? This is just a workaround but I find that it helps.

For sharpening I like to use MultiscaleMedianTransform on the linear integration with a lightness mask to protect the background. To check to see if the background is protected enough, I run the NoiseEvaluation script before and after the MMT. I like the background protection to be strong enough so that noise does not increase by more than a couple percent.

I have found that MMT itself can introduce dark pixel overshoot in certain situations, e.g. between two bright, closely spaced stars. So my MMT mask protects stars as well as the background. IMO any really high contrast edge needs extra protection.

I also just started an experiment with some other interpolation schemes. Currently registration increases the FWHM of my frames by a fair bit (15 percent or more). I hope to find something that controls FWHM better.

Regards,
Mike

[kwiechen]

Hi Andrea,

try 'Bicubic spline' with clamping threshold 0.1.

Kai

[lucchett]

Hi Mike & Kai,
thank you very much for your replies.

I will try Bicubic spline & clamping 0.1.
I will compare the results to the current Lanczos 3, 0.1 clamping I am using on a M33 I am processing at present.
With this settings I have been able to control ringing but stars exibith very thin "cross" that make them to seem not pefectly round. (I am using a refractor..)

Mike, I followed a similar approach (less sofisticated than yours) on my last image but I think that if we start with a sharper image the final result should be even better.
In my last Sh2-155 (www.andrealucchetti.smugmug.com) at the beginning I thougth the image was soft because of the hot & humid summer nights.
Now, I think I have better to re-process it, just to check!

Thank you again,
Andrea