Image of the day

Captured by
John Greenlee

M101 and others

My Account

New to Astromart?

Register an account...

Need Help?

reducing the martian hard edge

Started by awesley, 09/25/2005 11:21PM
Posted 09/25/2005 11:21PM Opening Post
I've been trying for the last day or so to find a way of processing mars that doesn't exaggerate the "hard edge" that we all seem to get, and I've made some progress.

Have a look at the attached image, again it's one from sept 24 but I don't have any new ones to try - it's been raining the last couple of days :-(

The edge is now a thin blue line, I think that's probably close to correct since there is an atmosphere and dust etc, we should see something around the sunlit side.

Now something else has shown up that's very interesting - look near the south pole and you'll see that the haze there is green, not blue. I hadn't seen this before, I think the true colour was lost in the overbright edge.

Overall I think the colour is better - Olympus Mons is now distinctly visible and it's a different shade of red to the surrounding plain.

Here's what I did, please try it and see if it helps your images:

- In registax wavelets, be gentle. You don't have to do all the sharpening in that one step, we can apply unsharp masks etc later on. Set layers 5 and 6 to -1 or maybe even less, this should help reduce the edge by subtracting off large, bright bits of the image.

- Next, save the image and load it into your favourite photoshop-like program. I use Astra Image, but I guess photoshop or similar would do as well.

- If you have a high-pass filter, try applying that. It will increase the contrast in the image without enhancing the edge.

- Apply an unsharp mask with a reasonably large radius, but only gently.

- Now smooth the image with something like an "alpha-trimmed mean" or "3x3 gaussian mean", whatever you've got. There will be artifacts introduced by the previous sharpening stages to get rid of, and this smoothing will help do that.

- Now apply another unsharp mask with a smaller radius to highlight the remaining features.

This is what works for me, but please remember that your mileage will vary, depending on your raw data.

I'd love to hear if this sort of approach can give you good images without exaggerating the hard edge.

regards, Anthony

Attached Image:

awesley's attachment for post 27077

http://www.acquerra.com.au/astro
Posted 09/27/2005 08:09AM #1
Anthony, that's a great image, and a nice step by step processing note... thank you! 8)

Ivan Gastaldo 8)
Coconut Creek, FL

Ivan's Observatory
Lat 26N 16' 48" Long 80W 10' 48"
[COLOR="Red"]Personal Website:[/COLOR] http://www.ivangastaldo.com

CCD Imaging and Processing/Deep Sky - Moderator
I like to complain about everything - Moderator
Posted 09/27/2005 02:57PM #2
Hi Anthony,

I've been playing with these directions and have a question or two.

1. In Registax, you set layers 5 and 6 to a negative value? Do you set the other layers as well? I tried this with an image I just processed and set layers 5 and 6 to -1 and 3 and 4 to +50.

2. I use Astra Image and really like it. I can't find the "high pass filter" you refer to. Does it have another name in AI?

3. AI only has strength and power for it's settings for unsharp mask, I'm not sure how to interpret your directions to the AI settings...would radius be strength or power?

4. I use the median filter for noise removal, is the median filter better or worse?

Thanks!!!
Paul

Anthony Wesley said:

I've been trying for the last day or so to find a way of processing mars that doesn't exaggerate the "hard edge" that we all seem to get, and I've made some progress.

Have a look at the attached image, again it's one from sept 24 but I don't have any new ones to try - it's been raining the last couple of days :-(

The edge is now a thin blue line, I think that's probably close to correct since there is an atmosphere and dust etc, we should see something around the sunlit side.

Now something else has shown up that's very interesting - look near the south pole and you'll see that the haze there is green, not blue. I hadn't seen this before, I think the true colour was lost in the overbright edge.

Overall I think the colour is better - Olympus Mons is now distinctly visible and it's a different shade of red to the surrounding plain.

Here's what I did, please try it and see if it helps your images:

- In registax wavelets, be gentle. You don't have to do all the sharpening in that one step, we can apply unsharp masks etc later on. Set layers 5 and 6 to -1 or maybe even less, this should help reduce the edge by subtracting off large, bright bits of the image.

- Next, save the image and load it into your favourite photoshop-like program. I use Astra Image, but I guess photoshop or similar would do as well.

- If you have a high-pass filter, try applying that. It will increase the contrast in the image without enhancing the edge.

- Apply an unsharp mask with a reasonably large radius, but only gently.

- Now smooth the image with something like an "alpha-trimmed mean" or "3x3 gaussian mean", whatever you've got. There will be artifacts introduced by the previous sharpening stages to get rid of, and this smoothing will help do that.

- Now apply another unsharp mask with a smaller radius to highlight the remaining features.

This is what works for me, but please remember that your mileage will vary, depending on your raw data.

I'd love to hear if this sort of approach can give you good images without exaggerating the hard edge.

regards, Anthony

Paul Maxson
Moderator - CCD Imaging and Processing/Solar System
Celestron C14, Mewlon 250, Orion EON120, ASI120MM, Flea3.
http://www.sunspot51.com