fixed pattern Noise reduction is it possible?

[gvanhau]

Hello

I have several sesions of images taken with my Atik 383L mono cam.
I found that the images were very noisy compared to my DSLR and my QHY8L and cloud not process them with good results.
Until a few weeks ago, I didnt know that this model of ATIK had problems when votage drops below 12.5V.

Now, looking to the individual captures, I can see that there is a pattern of noise that repeats from image to image even from sesion to sesion, the same noisy pixels apear on the images.

The noise is not removed by dark frames nor by flat or bias. (see image)

I tried using cosmetic correction, noise reduction, etc, etc, etc.
The last thing I tried was stacking the images wo aligning and the substract this result from the original images as a kind of artificial dark and the result was promising, but not good.

Do you have any processing suggestion for this?

Regards
Geert



 

[Alejandro Tombolini]

Hola Geert,

It seems a difficult task. Have you checked with Fourier if there is an identifiable pattern?

Saludos.
Alejandro.

[pfile]

can't this noise be rejected during stacking? or are the 'hot' pixels lining up on one another causing them to look like signal?

[gvanhau]

Hello
Thanks for your answers.

Hola Geert,

It seems a difficult task. Have you checked with Fourier if there is an identifiable pattern?

Saludos.
Alejandro.

No, I didn't, I doesn't seem an geometrical or periodic pattern just the same pixels (I think random layout) show up at the same position on every frame, however, I will give it a try, may be there is a pattern that I don't see .

can't this noise be rejected during stacking? or are the 'hot' pixels lining up on one another causing them to look like signal?

Well, pixels are always at the same place of every frame, but because of drift they build up like a brush strike during stacking.  I played a lot with the different rejection algorithms and  forced very strict rejections but  even then bad pixels aren't completely removed. 

Regards
Geert

[Geoff]

You may not like this answer, but the second post in this thread
http://pixinsight.com/forum/index.php?topic=4075.0
will tell you what to do.
Geoff

[gvanhau]

Thaks Geoff
I was thinking the same, but I wont give up so easily ...

Regards
Geert

[martin_magnan]

Geert

This pattern is often caused by flexure in your optical setup and it is accentuated by calibration with not enough calibration frames (like 10 bias or 10 dark etc...)

Bad polar alignment will also cause this kind of pattern but with a circular aspect (rotation).

My recommandation

-fight flexure if you guide
-use lot of calibration frames (100 +)
-dither between light frame

Martin

[gvanhau]

Thanks Martin

Source of noise is already identified: when Cam votage drops below ~12.5V this noise comes up.
I didn't know this during those sesions, and was working at field with a 12V battery.
I'm now working with a "step up" regulator that ensures the right voltage.

What I am now looking for if there is any alternative to recue those imaging sessions.
Dark and bias frames dont work, because that noise doesn't show up in them, it is as if those pixels are more sensitive to light when votage is low, but dark an bias frames are taken w/o light.  I tried to generate images at very low light, as to simulate the same condition of the images w/o results until now.

Regards
Geert

[niteman1946]

Source of noise is already identified: when Cam votage drops below ~12.5V this noise comes up.
I didn't know this during those sesions, and was working at field with a 12V battery.
I'm now working with a "step up" regulator that ensures the right voltage.

Hi Geert,

If I can ask more about the camera voltage .....
I'm considering going with the Atik 383L+ mono (same as yours, I think).  What voltage do you consider safely minimum?  And how did you get there?
Thanks,

Mark

[gvanhau]

Hi Geert,

If I can ask more about the camera voltage .....
I'm considering going with the Atik 383L+ mono (same as yours, I think).  What voltage do you consider safely minimum?  And how did you get there?
Thanks,

Mark

Hello Mark
I am not realy shure what the real minimum is, I supose it can be CAM dependant. 12.5V seems to work well for me and for other people on the internet.   
I currently use a 100W alternative power supply intended for notebooks. This one has dual input (mains and a 12V) and a switchable output from 12 to 24V.
I found out that the output  at the 12V setting really was 12.5V even if the input was as low as 9V.   I tested a seccond power supply from a  different brand and it behaved similar.

Regards
Geert

[martin_magnan]

I See

With the proper voltage, the north-south streak pattern disappear ?

Martin

[niteman1946]

Thanks Geert,

Mark

[Ignacio]

Geert,

It seems that what you need is a superflat, built with frames taken at the same operating voltage.

Ignacio

[gvanhau]

I See

With the proper voltage, the north-south streak pattern disappear ?

Martin

Well, it should, if the noise on each frame was random...

Geert,

It seems that what you need is a superflat, built with frames taken at the same operating voltage.

Ignacio

Thanks Ignacio. What do you mean with superflat? I have flat frames at the same voltage, but exposures are much shorter and if noise pixels are there, they are not visible because the whole frame is lighted.   What I think could help and may be the same you are meaning is to take sky images at random positions and then stack them ?


Regards
Geert

[Ignacio]

That's excactyl what a super flat is. Take a look here http://www.astro.uni-bonn.de/~mischa/datareduction/superflat.html

best
Ignacio