PixInsight Forum

PixInsight => Announcements => Topic started by: Carlos Milovic on 2014 November 17 12:00:35

Title: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2014 November 17 12:00:35
Dear all,

We are starting a new processing tool, for automatic alignment of planetary, lunar and solar images, using the same engine that StarAlignment has (handle distortions, etc). This would be compatible with drizzle, of course.

Right now we need small sets of samples (no more than 50 frames) of some typical lunar, solar and planetary targets. If there are large distortions due to atmospheric turbulence, it wold be great.

We plan to release a first version of this tool the first days of December, so we need your help with these small samples to create the core feature extraction algorithm, and test the feature descriptors used in the sample matching.


Thank you very much for sharing your data with us!
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Zocky on 2014 November 17 12:57:29
I have CR2 shots of a Moon with Canon EOS 50D. Is that also OK? If so, where can I upload them and in what resolution?
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Nocturnal on 2014 November 17 13:36:16
I have some pretty cr@ppy data you're welcome to break your teeth on :) It is in SER format. Is that OK? You can also see what AS2! made of it.

https://drive.google.com/open?id=0B2nw0jiNO0X8bjFCODVYb1o3MmM&authuser=0
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2014 November 17 13:42:47
Yes, indeed they should work. Thank you, Zocky. You may upload them to the Endor server, or share with me a dropbox link. Tell me if that is fine with you.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Zocky on 2014 November 17 13:47:40
I will upload them to the Endor server. Thnx.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2014 November 17 13:53:31
Thanks, Sander. I'm already downloading the file. I'll let you know if there is a problem with it.

Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Zocky on 2014 November 17 14:30:29
There are 2 foders on Endor. Moon_Zoran_Novak_Zocky and Sun_Zoran_Novak_Zocky. With CR2 files I have included my end results, made with other softwares. Files are not perfect, but for testing purposes I hope that will be even better.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: chris.bailey on 2014 November 18 00:54:10
Carlos

I have put some mono fits files in a folder on Endor (Moon_cbailey) which are in four mosaic parts. Hopefully the file names make sense.

Chris
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: pfile on 2014 November 18 10:24:24
here are 5 zipped .avi files of the moon. lots of atmospheric distortions :(

https://drive.google.com/folderview?id=0B6cW4ae2KyMJRXZyMTBjVUotb00&usp=sharing

moontest ... is from a skyris video camera, the others are from a DSLR running live view capture mode. hopefully the avi is readable by your software. should just be MJPEG codec, if i remember right.

rob

Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2014 November 18 10:35:57
Thanks Zocky and Chris, I already downloaded your files. :)

Sander, I'm having problems with your SER file... ffmpeg doesn't support it. How do you convert that to another lossless format?

Thanks Rob. Downloading your data too :)


I think that with your data, guys, I have enough for the design of the feature detector. Thank you very much! I'll let you know the news of the development of the new tool. :)


Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Zocky on 2014 November 18 13:08:23
Carlos, this file (https://drive.google.com/file/d/0B-rFsWOBGhrURjNtVjNyS3BaRVU/view?usp=sharing) is too big for Endor. It is an avi file of the Sunspots, made with DMK camera and Lunt Ha solar telescope.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: mschuster on 2014 November 24 11:16:11
Carlos,

This maybe is something your new process could handle:

Here is a zip of defocused bright star images, 16 inside of focus frames and 16 outside of focus frames. Fresnel rings are visible, somewhat distorted by seeing.

https://dl.dropboxusercontent.com/u/109232477/PixInsight/DefocusedStar.zip (https://dl.dropboxusercontent.com/u/109232477/PixInsight/DefocusedStar.zip)

Currently I threshold, find the barycenter of the binary image, and then cocenter barycenters via simple translation, bilinear interpolation only. Ideally the registration should mitigate seeing distortions and also ideally preserve total flux. Here are example integrations.

(http://mschuster.zenfolio.com/img/s8/v75/p199886793.png)

FYI, integrations are input to my Fourier wavefront analyzer which measures optical aberrations of my lens.

http://mschuster.zenfolio.com/blog/2013/12/on-axis-image-quality (http://mschuster.zenfolio.com/blog/2013/12/on-axis-image-quality)

Thanks,
Mike
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2014 November 24 11:21:25
Thanks Zocky for that last example!

Mike, I'll take a look at your images as soon as I have the first prototype working. Looks like an interesting application. :)

Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Torsinadoc on 2014 November 26 13:02:51
i have some white light full solar disc images if you need them
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: james7 on 2015 February 16 03:11:36
Is there anything new on this tool?

If you are still looking for sample images I can provide some for several different targets.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2015 February 16 06:10:21
Hi James

We have written a first version of this tool, with a few algorithms that we are trying. Unfortunatelly, it does not yet meet our performance standards, so it needs a lot of work left). We are also planning to include the SURF algorithm, so this tool could be used with common daylight images too.

On the other hand, this month we started a side proyect, the planetary subframe selector. We completed a first study of over 300 quality estimators, using a frame sequence of Jupiter. We have chosen the best 15 to build a draft tool that should be ready in the next weeks for all of you to try.
To make this study better we are in need of frames that are manually scored. Alejandro and Enzo kindly scored 600 frames of jupiter (integer values from 1 [worse] to 5 [better]). We also need similar samples for other targets. I know that this is a lot of work, but his should provide with a set of quality estimators that are good enough for a variaty of targets and sky conditions, that allows us to build a suitable clasificator.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: james7 on 2015 February 17 02:45:48
Okay, thanks for the update.

I have lunar images and some of Jupiter and even some stellar/DSOs that I've graded by "hand." The reason I have these is that I sometimes do high-resolution work a little differently than most. I capture using an APS-C camera in STILL mode (not video) and take anywhere from 48 to 64 individual, sixteen-megapixel stills and then visually inspect those to find the best given the seeing and other factors. I generally concentrate only on a relatively small area of interest (craterlet or rille on the moon, cloud detail in Jupiter, double star, etc.) and typically reject anywhere from one third to one half of the candidates (depending upon the seeing, which generally must be pretty good or why else bother?). On the lunar images, if it is obvious that only a small portion of the image is sharp then I will reject the sub even if it looks good in the primary area of interest.

After that process I run the images though either Registax or AutoStakkert!2 and then sharpen in either Registax (wavelets) or PixInsight (deconvolution).

Honestly, however, for planetary images video is best because you obviously need hundreds if not thousands of "good" frames to get the maximum detail possible and I would never be able to grade that many images by simple visual inspection (other than to do a quick, gross inspection to reject the very worst images). Thus, when capturing with my dedicated planetary camera (video) I almost always allow Registrax or AutoStakkert! to do the automatic selection.

However, for lunar images I think the situation can be a little different since I can cover much larger fields with an APS-C camera than I can with the typical one or two megapixel planetary camera. That's really the major reason why I've continued with this technique (i.e. stills rather that video), because I like the wide field of coverage I can get with the APS-C sensor when I image the moon (rather than having to do a multi-framing montage to cover that same area with a small-sensor camera).

I'd be willing to provide you with some sets of images on the moon. However, unless your tool can easily handle four megapixel or larger images then my workflow probably wouldn't benefit much from what I can currently do with either Registax or AutoStakkert!. The problem with those tools is that they can't handle large images which to some degree has limited what I can do with my APS-C sized frames.

Let me know if I can help in any way, as I'd absolutely love to have a tool that could handle the grading and integration of really large (greater than 4 megapixel) lunar images.

Here are some links to two lunar shots that I've done using stills from my APS-C camera (single framing, not a montage, these are large images so make certain you either zoom or look at the full-size version, not just the default view in Flickr):

  https://flic.kr/p/pf7Trc

  https://flic.kr/p/qhrn3w

 
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2015 February 17 07:38:10
Hi James

Certainly this new tool should work with large images, and large data sets. I am writiing right now the source code that analyses a single image, and extracts a vector of features, that later will be used by a script to make the sorting and selection.
If you have those manually graded frames, I'd love to run the analysis of relevant features with them too. I'll need a csv or txt file with all the grades and filenames, to asign quality classes in the analysis.

Thank you very much!
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: james7 on 2015 February 17 19:57:19
My images haven't been graded on any scale other that an overall accept and reject. Further, on the lunar shots the seeing conditions vary so much over the full field of view that it would be VERY difficult to weight the quality of the entire frame by visual inspection alone. I'd pretty much have to divide the image into a large matrix and grade each area separately and then sum the individual elements (maybe scoring a one for an accepted zone and a zero for a reject). The other issue is that when working with such large fields and at such high magnifications I've yet to find an arrangement that will produce a flat field (in resolution) over the entire APS-C frame. It wouldn't be that much of an issue to crop only to the "good" part of the field but unfortunately when tracking the moon I've never gotten the image to remain that well centered over the four or five minutes of capture times that I typically use. Thus, each crop would have to be a little different if I wanted to maintain as much of the field as possible.

I could give such a process a try, but I suspect that I could only do a few frames that way since it could probably take many hours to do that type of grading on the full data sets that I have used in the past.

If it takes me a week or two to find the time to do maybe five to ten total frames would that be good enough for your purposes?
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2015 February 18 07:36:50
Ok. Don't worry, then. I need large samples (dozens or hundreds) of the same field/target to compare the quality stimators.
We could follow a sliding window scheme for grading those large images, but performing the manual job would be too much. I think that at this stage it is better to stay with narrower fields.

Thanks anyway :)

Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: slang on 2015 March 31 04:21:56
Indeed, this tool sounds quite fantastic!

I have some old avi files with small images (webcam) in them (to make up a mosaic), as well as some more recent (~50 images or so) DSLR shots (single frame, no mosaic).

Cant wait until the tool is released and I can try it out :-)

I don't have any large sets of manually sorted images though, sorry...
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: pengsloth on 2015 April 07 14:19:13
I have a 2GB video (1802 frames) of Saturn at 0.084" per pixel. How could I get it to you?
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2015 April 08 12:41:30
Sounds excelent. I took a video of saturn with my dslr, but the quality is awfull, so yours is a blessing!
I think that the better way is the Endor server. http://endor.uv.es/ If you don't have an account, ask for one.
You may also share a Dropbox folder, or in Google Drive...

BTW, I'm a bit stalled with this project. I have a working module that calculates some candidate quality estimators, but I need to write a script to make a UI for inspecting all the frames. Once I figure this out, things should speed up (in terms of news) a lot.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: pengsloth on 2015 April 09 00:10:45
Ok here is the AVI file. I also have a set of BMP frames (I no longer have the .MOV file from the DSLR that made it) for Jupiter a few months back when Io, Europa, and Ganymede were transiting it.

Saturn: https://drive.google.com/file/d/0BzoxZr4z_GA3Nm1tNm82MkR1X1U/view?usp=sharing

Let me know if that link works.
Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Carlos Milovic on 2015 April 10 10:19:46
Thanks! It took a while to download, but works fine. :)

Title: Re: Call for images: Planetary, Lunar and Solar images
Post by: Solidarity on 2015 August 02 23:04:01
Hello Carlos,

looking forward to the script. I have 100 RAW Files of a daylight moon. If it would help I can do an upload.

http://pixinsight.com/forum/index.php?topic=8758.0 (http://pixinsight.com/forum/index.php?topic=8758.0)

Karsten