Gaussian Deconvolution

[MarkS]

For the last 18 months or so, I've been working on a project that makes true Gaussian Deconvolution really easy to use, even for someone who's never even heard of it - in fact, to make it even easier to use than Wavelets.  It builds upon a (non-astronomical) project I've been working on for over 10 years to refocus out of focus photos.

Mathematically, deconvolution ought to give superior results to "Wavelets" for astronomical images but it tends to be more difficult to use.  However, I've now got my program to the point where in just a few seconds it produces a much better result than, for instance, using the Wavelets function in Registax.  And the beauty is you don't even have to know what you are doing - just play with two slider controls until it "looks right".  The aim is to produce sharp, "natural looking" lunar and planetary images without the obvious signs of overprocessing that are typically seen. In addition an "expert" user can go even further and perform direct image manipulation in the wavenumber (i.e the frequency or Fourier) domain if they so wish.

My recent Shickard image was processed using it:


I hope to to host an imaging session sometime in the New Year to explain the concepts behind it and to demonstrate it in action.  Also to make it available to anyone who wants it - the program should be stable enough by then.  It will be free of charge since it is still really only a prototype - I simply don't have the time to write, maintain and support a production quality application.

However, in the meantime I'd like some help.  I need more test data - stacked Lunar or Planetary images but without any kind of further processing such as Wavelets.  If you have an image or two that you think could help then please send them to me in a lossless (.bmp or .tiff but not .jpg) format.  Ideally the image should have been saved directly from the stacking program (e.g. Registax) in that same lossless format.

Thanks,

Mark

[Mike]

Nice one Mark. The algorithm might be marketable so I would keep it under wraps if I were you, maybe just release it to those you trust.

[Fay]

Well done, Mark, look forward to giving it a try.

I have not saved any unprocessed Moon images, but will do so in the future, for you.
Hopefully someone else, who has done more moon & planetary images, will be able to give you some to practice on.   

[Mike]

I'm sure I have plenty from my early days somewhere. I will take a look after the holidays.

[RobertM]

Sounds like a very interesting project Mark and as Mike says, definitely marketable.

I have a number of unprocessed moon images using the H9 which you're more than welcome to.  Do you want them stacked first or just individually converted to tiff ? 

[MarkS]

Thanks Robert.  Please stack them first.

Mark

[RobertM]

Where do you want them ?

[JohnP]

Mark - I save all my avi's & so def have some early moon efforts...

[Carole]

Sounds great, I think I have some unprocessed AVIs I could stack and send you.

I'll have a check after Xmas.

Carole

[Rocket Pooch]

Mark,

What development tools are you using for the processing?

Chris

[MarkS]

Microsoft VC++ with MFC for the Windows API.

[Rocket Pooch]

ah ha, ok wondered, I'm not going there again, do you know if there are any libraries for C# which read FIT and AVI Streams, I've not been able to find them on MSDN easily

[MarkS]

Have a look for AVIFileOpen etc.   There are a whole load of AVI functions - I think there are C# versions of them.

Dunno about FIT though.

Mark

[Rocket Pooch]

ta

[MarkS]

Recent cloudy and misty nights have meant that I have made faster progress than expected on this project  
Tonight I'm writing a brief tutorial to the program.  I should be able to post it up tomorrow with one or two sample images.

The algorithms used are all very standard - there's nothing proprietary or special about them.  I've just made them (hopefully) easy to use by having a simple interface that a complete novice can use.

Mark