Problem with MMT and dark pixels

[DanielF]

Hi.
I'm trying to do some noise reduction with the MultiScaleMedianTransform module but I'm getting some strange results.

My image has a lot of noise and after applying MMT some darker pixel are unaffected by the processing and it doesn't look so good.
I've used a clone of the image as a mask for the high signal areas.

I have attached images showing my image (a preview) before and after applying MMT, and the MMT-settings.

Anybody that has any idea what I'm doing wrong or if the problem is due to bad integration or something else?

Regards
Daniel

[NGC7789]

I'm still a newbie but I believe the "adaptive" control takes care of this.

[DanielF]

I'm still a newbie but I believe the "adaptive" control takes care of this.

Thanks! I actually tried different settings on the adaptive control but I apparently used too low values. It got better when I cranked it up a bit...

[topboxman]

I use the same MMT settings as yours plus I add to Layer 4:

t = 0.5 and s = 0.66

Try that and see if that helps.

I use MMT with pretty good results. I also clone the image and apply as an inverse mask.

Peter

[marekc]

I think the recommendations to play with the `Adaptive' settings may be correct. I managed to eliminate some dark blobs after MMT processing, by playing with the Adaptive settings:

http://photonshotnoise.blogspot.com/2012/07/making-adaptive-progress-with.html

- Marek

[topboxman]

Very interesting. I will have to try that. Are there any negative side effects?

Thanks,
Peter

[marekc]

I didn't see any obvious side-effects. In the case of the Virgo image I was processing, I found that the right Adaptive settings managed to get rid of the `dark blobs'. If this created any new artifacts, I didn't notice them.

[topboxman]

I tried this method last night. I noticed a small side effect. It seems to over smooth the image. It connects stars that are close together and draws white lines in between stars only if the stars are close together.

Peter