Trouble Extracting Small Scale Features

[andyschlei]

I am working on an image of NCG 2903, using the tutorial on multiscale processing.

I am trying to enhance my luminance image.

I have followed the instructions through the isolation of the large scale structures.  When I try to get to the small scale structures by subtracting the large scale from the base image, I get an unusable result.  I can't figure out what I am doing wrong.

The result of the subtraction has good small scale results in the galaxy, but the background is washed out and as bright as the galaxy area.

My large scale structures are at the 32-bit level, given the size of the object in the image.  It was imaged with a C-11 @ f5.95 (fl = 1666) and about .8 arcsec/pixel.

The bad resultant image is shown below.

The source FITS files for both the luminance and large scale are at ftp://ftp.obsballona.net/
username: anonymous (at) obsballona.net
password: your e-mail



Thanks in advance for the help.

--Andy

[Juan Conejero]

Hi Andy,

Sorry for the long delay in my answer.

I can't download the FITS files. When I login as anonymous user, the /pub directory is empty. I need to see the images to figure out what's wrong with your processing.

On the image you posted here, those rings around the stars look a bit weird. There are bright rings around the dark rings, indicating a possible wrong subtraction of the large-scale image. Did you subtract with Rescale enabled in PixelMath, when rescaling shouldn't be used? Just an idea, but we need to have a look at the FITS files.

Thank you

[andyschlei]

Juan,

No problem on the delay, I really appreciate the help.

I am having problems with my hosting provider regarding anonymous FTP .  If you have an FTP client, use the user name "anonymous" at "obsballona.net" replacing at wth "@" and you should be able to get to the files.

I have also put them up to be available in HTTP.  Try these links using a "save link as" approach.

Large Scale

Base Image

I did rescale during pixel math.  I found that I had the same issue but at lower overall pixed values when I did not scale.

Thanks!

--Andy

[Juan Conejero]

Hi Andy

I think you'll be interested in trying out our new HDRWaveletTransform process on your image. I have made a quick test:

Your original luminance image:
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/01.jpg

Processed with HDRWaveletTransform (parameters as shown):
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/02.jpg

This process performs extremely well with all kind of deep-sky images. See how contrast has been enhanced locally just where the image needs it. Playing with the number of layers you can control the scales of the structures that the algorithm mainly works on.

Once the image has been flattened, a simple CurvesTransform can be used to modify the global brightness/contrast ratio:
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/03.jpg

The curve shown is just one among many possibilities. This is where your personal preferences and taste enter the game.

Then, noise reduction:
Before ACDNR:
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/04.jpg
After ACDNR:
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/05.jpg

And finally, HistogramTransform:
http://pleiades-astrophoto.com/examples/HDRWT/NGC2903/06.jpg

This is just a quick test. In particular, the stars have not been correctly processed. A careful star protection mask should be used to protect them from bloating during HDRWaveletTransform. But that's the idea ;-)

Hope this helps

[andyschlei]

Juan,

This looks great.  Thank you so much for the examples!

I've been travelling and away from my images, so I am primed to try out the new tool.

I have struggled a bit in the past when I have tried to create a star mask.  I seem to have the star mask too small, so I still get artifacts around the stars.  Could I use a mask that only reveals the galaxy?  Or is there a good approach to creating a star mask that I should use?

Thanks again,

--Andy