ringing artefacts in lunar images

[Shiraz (Ray)]

Took some lunar images recently and noticed (again) slight ringing artefacts in the dark regions of craters (see example attached), where there is a thin white line inside the core region and the central dark part is not black. The dark crater and rille regions also throw slight shadow lines into the brighter zones and the small craters look more like pimples than depressions.

In an attempt to understand what is going on, I made the synthetic black and white image shown as the upper section of the insert in the image below and;
1. convolved it with an "Aberrator" point spread function for my scope to simulate the image formation process
2. sharpened the resulting blurred image using a VanCittert deconvolution. this algorithm is typical of sharpening operators in that it boosts image content at high spatial frequencies.

The result is shown in the lower section of the insert (with dynamic range reduced) - all of the features of the real image are reproduced (I didn't adjust the scales yet), from the the thin white lines in the dark regions to the lightening of the central zone and the external dark lines. I could not get this result with any other approach.

Thus, it seems that the artefacts are a normal outcome of applying sharpening to images that have been blurred by an Airy pattern. Probably have to live with this, but there may be implications for optical design and quality - eg do SCTs or mirrors with low Strehl, which both have stronger Airy ring structures, produce more pronounced artefacts? And how much damage does this do in lower contrast images where there are fewer step changes in intensity? ...And maybe it is possible to tailor a sharpening algorithm to take account of the real PSF of the scope that took the image. Food for more thought. regards Ray

[Troy]

Hello Ray, the real image looks very different to me. I don't think I have seen this on the moon before.

Do you get any ringing affect when imaging planets?
Is the artifact in the raw images?
What are some image details e.g file type captured with, frame rate and camera settings?

Did you use the imaging source camera? Which are known for the ringing artifacts and there is a lot of info on the net about this.

There is a lack of contrast between the bright and dark areas(sudden cut off to pitch black) very strange.

[Shiraz (Ray)]

hi Troy.

Thanks for your interest. As far as I can tell, this effect is widespread in lunar images taken in good seeing and with low angle illumination. There is often that little white line tracing around the boundary inside the dark zone of high contrast craters. A quick look around found a few examples from a variety of operators, cameras, optics etc (see especially the craters in the upper left of the second listed image):
http://www.damianpeach.com/images/lunar0709/baco_2007_05_23dp.jpg
http://www.damianpeach.com/images/lunar0709/beaumont_2007_05_22dp.jpg
http://the-moon.wikispaces.com/LPOD+May+28%2C+2008
http://the-moon.wikispaces.com/LPOD+Apr+15%2C+2008
http://lpod.wikispaces.com/May+13%2C+2011

All of my images were taken at 60hz with a DMK618, which does not have the dreaded 60hz double image artefact - I also have an 098 based DMK camera, but that does have the artefact and I don't use it much anymore for that reason.

The simulation was intended to show that the same little white line can be produced by sharpening a synthetic image that started out life with an abrupt high contrast edge and then was blurred by an Airy pattern. I don't think that the effect has anything to do with the camera - the simulation was prepared without using one and suggests that the artefact is an unavoidable outcome of the optics PSF and sharpening combination. The simulation also shows the more subtle dark boundary echo that can be formed in the bright zones of such an image and this effect can also be seen in the real images.

There are a couple of ways to mitigate the "white line" effect. One is to use "curves" to black out the darker areas of an image and many published lunar images look to have been processed this way - I guess it would also be fairly easy to manually brush out the effect in an important image. The other is to image with more direct lighting, so that there are no extreme contrast regions in the image - many lunar imagers do this.

regards Ray

[Troy]

Hello Ray, very interesting.
I'm not totally convinced it's a sharpening artifact(I know you did a simulation but hmm).
Have you tried a apodizing mask?
To see if the white line disappears when the mask has been used. This would try and negate the effect of the airy disc pattern. 8-)

[cybereye (Mario)]

Ray,

Are your tests just on decon or have you tried wavelets as well?

Cheers,
Mario

[riklaunim]

That's a diffraction pattern that will show up at high resolution - Airy disk. Oversharpening will make it clearly visible. The same thing is responsible for bad edges on Mars or Venus.

[Paul Haese]

This is a common problem in SCT's with less than perfect seeing. As the seeing degrades the effect becomes more pronounced. Edge sharpening tends to enhance the problem more, so careful sharpening via masks can be a better way to deal with the problem.

[Shiraz (Ray)]

Originally Posted by Troy

Hello Ray, very interesting.
I'm not totally convinced it's a sharpening artifact(I know you did a simulation but hmm).
Have you tried a apodizing mask?
To see if the white line disappears when the mask has been used. This would try and negate the effect of the airy disc pattern. 8-)

Hi Troy. The secondary ring appears at about 0.5 arc sec from the main edge, so I think it is a good candidate to be a sharpened diffraction ring, as per the simulation. Apodising mask is a great idea - will try it at some future date.

Originally Posted by cybereye

Ray,

Are your tests just on decon or have you tried wavelets as well?

Cheers,
Mario

Hi Mario, basically the same effect with wavelets, deconvolution, unsharp mask....

Originally Posted by riklaunim

That's a diffraction pattern that will show up at high resolution - Airy disk. Oversharpening will make it clearly visible. The same thing is responsible for bad edges on Mars or Venus.

hi Piotr. agree - have seen a few bad edges on mars this season

Originally Posted by Paul Haese

This is a common problem in SCT's with less than perfect seeing. As the seeing degrades the effect becomes more pronounced. Edge sharpening tends to enhance the problem more, so careful sharpening via masks can be a better way to deal with the problem.

I guess SCTs have stronger ring structures, so these observations seem to reinforce the idea that this is a result of the Airy pattern.

The basic problem I have is that the image on the screen is a sampled representation of the original image blurred by a variety of atmospheric effects (probably a Gaussian blur is a fair representation of these), the optics PSF (definitely not a Gaussian) and the camera diffusion spread function (again Gaussian may be reasonable). The sharpening algorithms (as far as I know) assume a smooth blurring function and try to correct for it - amplifying artefacts from the optics PSF in the process. This just seems like a dumb way to do it, since it would surely make more sense to correct for a real blur function than an unrealistic assumed one.

What I think might be possible is to find the actual blur function by imaging a star as well as the target and using the star shape as the kernel for the sharpening alogrithm (some of the available algorithms allow arbitrary kernels) - have used with success on DSO images, but not yet on solar system stuff. Has anyone ever tried this?

regards Ray