Printable Version of Topic

Posted by: Ant103 Oct 17 2006, 08:52 AM

Hi,

Here is a try to super resolution imaging, by using http://www.astrosurf.com/buil/us/iris/iris.htm, a software made for astronomical imagery processing.

I took the http://origin.mars5.jpl.nasa.gov/gallery/all/spirit_p984.html and stack and align them to have a better resolution.

Before, I resised all the frames at 200%.

One of the 15 originals raw images :

And the final picture :
http://img100.imageshack.us/img100/6208/sol984superres15img200p100is0.jpg

The result is a little visible, but, to have better resolution, I need more than 15 pictures (in astronomical processing, we need a thousand pictures extract from an avi film).

Posted by: djellison Oct 17 2006, 09:00 AM

Realistically, you need to be using the uncompressed data - not raw JPG's complete with compression artifacts.

The analogy with planetary imaging with a webcam doesn't hold too well. The reason one takes hundreds of images is because within those hundreds there will be perhaps a dozen frames of near perfect 'seing' where just for a moment the column mass of air was not screwing the image quite so much.

With surface-to-surface imaging such as these super-res sequences, that's not the case - every frame you aquire is a 'good' one so you don't need hundreds of images.

What might be interesting is to compare one of these sequences ( given that the early Olympia outcrop super reses from the early 700's is on the PDS now ) - and try it with the JPG's - the RAD's and then compare to the Pancam teams efforts - http://pancam.astro.cornell.edu/pancam_instrument/superres.html

Doug

Posted by: Ant103 Oct 17 2006, 10:28 AM

Yes, I know this fact and know the difference between jpl-exploratorium pics and the original calibrated raws.

Here is my try on the Sol 899 images :

You can now compare it to the Cornell SuperRes Project . There is a difference isnt'it ?

Posted by: tedstryk Oct 17 2006, 10:39 AM

I have played around with the JPEGs, and yes, super resolution processing does little. But I have worked with some early mission raws, and there is a lot to be gained.

Posted by: ugordan Oct 17 2006, 10:46 AM

Are you applying any sharpening to the result? These images look pretty sharpened -- maybe an artifact of Iris' processing?

Posted by: Ant103 Oct 17 2006, 10:52 AM

Yes, I apply a sharpenning to the result and some artefacts appears. These are due to Iris process.

Posted by: antoniseb Oct 17 2006, 11:41 AM

With most astronomical images, there is very little change from frame to frame, but in this case you need to be dealing with the changes in the shadows too. That must make a strange distortion on your end product.

Posted by: MarkL Oct 17 2006, 05:25 PM

There's noticeable sharpening in the stacked image Ant. What this type of stacking does is improve signal to noise and seems to have done that in Ant's version.