Asteroid Zoo Talk

A Sneak Peak

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    We’ve been working on getting your results packaged – and understanding where you (the AsteroidZoo community) excel vs. the machines.
    Here’s a good example – this is a trimmed area near a star – and you found an asteroid that was totally missed by all the software and didn’t seem to be in the catalog (the Minor Planet Center will be the final check on if this is a known object that now has a better orbit or if it’s something that’s a new find).
    Updates coming shortly.

    enter image description here

    Posted

  • Barbalbero by Barbalbero

    Thanks very much for the reply.
    I remember the first possible asteroid I observed with this project was a case similar to this one.
    I cannot wait to have news about our discoveries!!!

    Posted

  • DZM by DZM admin

    Hm, typically Talk auto-condenses those images. I have no idea why it didn't here.

    If you want to see the full image, click here: https://dailyzooniverse.files.wordpress.com/2015/01/star-grazing_azoo.png

    Posted

  • Andy_Arg by Andy_Arg

    here is the image set that we saw
    http://talk.asteroidzoo.org/#/subjects/AAZ0002ul8

    Posted

  • grums by grums

    I'm surprised that the software didn't pick that one up. It must be in the top 1 or 2% in terms of ease of spotting visually. What is it that makes this asteroid difficult to auto-detect within the set?

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    This particular asteroid is too close to the stars. The bright one is saturated (the black/white in the center) and likely this general area was eliminated as "bad" because of that.

    This is a spectacular highlight of where you the folks are better than the machines. You can use judgement to say "yeah - there's a star right there, but that spot corresponds to that other spot".

    Again - many thanks for your hard work.

    Posted

  • grums by grums

    Thanks Dr A. Perhaps not a difficult fix for the software though?? BTW are these images filtered to reduce bandwidth and spread the noise where the originals would have had more pixels and, possibly, more brightness levels? I'm just curious, as I could see this would be useful and also read that the CCD arrays were larger than 256x256 though these images may not be from the whole array. I guess there maybe some post processing for aperture correction too?

    Posted

  • Barbalbero by Barbalbero in response to Dr.Asteroid's comment.

    Thanks very much for your recent information, Dr. Asteroid.
    Just for curiosity, how many images have been sent to the Minor Planet Center in order to have a final check on if we observed a known or unknown object in the sets we studied? How long does it usually take in order to have a reply from the Minor Planet Center?
    I will continue for sure to analyze images here on Asteroid Zoo, hoping to give a good contribution and making new discoveries

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    We've had some setbacks in terms of getting the pipeline operating correctly. We're being careful about just dumping a vast number of asteroids from archival data (we're working with it).

    So far, 35 have been done as a run. Results are not back yet. I will update when they are.

    Posted

  • hightower73 by hightower73

    So what has to be done to verify the images, is there any thing we can do to help ???

    Posted

  • Barbalbero by Barbalbero in response to Dr.Asteroid's comment.

    Thanks very much for the information.
    I studied many images for this project, and I am sure there are many interesting objects which are good candidates. Other times it is very difficult to understand if the objects with apparent movements are real asteroids or just artifacts or "problems" in the images, so it is right you need to pay attention to the set of images you send to the Minor Planet Center. I wish there will be many discoveries thanks to this project, if there is something I can do (I mean other than studying the images) in order to help the project just let me know it. Maybe with the collaborations of people like me involved in this project as hunters, the Administrators of the project can have a good contribution to their work (I know it is a very difficult work, for this reason it can be helpful to involve us).

    Posted

  • grums by grums

    I have to say I have learnt a huge amount about the problems of finding asteroids from this activity because I have had to do some further research and think about the problems involved. The whole concept of combining a privately funded business with, what is a relatively poorly funded, international academic search to identify potentially dangerous NEAs is quite enterprising too. Could be a sort of win-win situation though must be difficult to balance at times.

    Probably the biggest practical problem is image quality and consistancy as has been alluded to above and on other occasions. I can see the images are what is currently available so there is no choice. I can also see that the best way to progress asteroid searching is to do it from space to avoid all the atmospheric issues. Even a relatively small aperture telescope would be a big improvement, although given the costs vs payload it would be a shame to not do as well as can be achieved to maximise resolution. To do the job properly via optical search I would suspect that at least 2 satellites would be good plus a fixed station on Mars to detect asteroids approaching from the sun 😃. Wide baseline Radar looks promising.

    As far as searching the current images, I for one, would like to understand the procedures used to verify the identifications. The 4 closely spaced dots are probably not sufficient to present the the Mnor Planet Center so I assume that you must use the approximate vector, calculated from the identified 4 dots, to search a contemporaneous image set where the asteroid is predicted to be, and confirm (or not) that it is there. And to get an orbit you would need a further image set. Is this right or, if not, how do you do it?

    Also, I asked this before, are the 256x256 images part of a larger single image (1024x1024 say), are they all that is captured (i.e.on a 256x256 CCD array) or are the the result of some filtering of data from a larger array?

    Why is there frequently 1 or more pixel of jitter between frames? I feel this could be reduced and would help both automated and manual detection. If the non-moving images remain in the same place they can be removed by pre-processing for example. This still leaves a vast array of other image issues to resolve however. Having played a bit with the image sets I can see the biggest problem is finding any consistant methodolgy to detect an asteroid because of the image variability. The best way is probably to identify and eliminate each image problem first and then process the resulting, more consistant, image set to identify set of 4 moving group of pixels and then calculate a vector from them. This is all quite compute intensive but with a lot of people with powerful PCs and suitable software I feel this could be an alternative approach. Even just providing a tool set to manipulate the image sets may be helpful - it is a very time consuming process doing it with Paint Shop Pro or Photoshop and no way to get the images back to the 4 image set. Anyway, I don't suppose you guys have the effort to do much, if any, of this. It would just be a nice-to-have.

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    Thank you all for your continued interest.

    grums - sorry for not catching your earlier question:

    The Minor Planet Center (MPC) requires centroids in RA/Dec of at least three points. They compute the vector and impose that on the system with their own modeling software. If this corresponds to a physically plausible orbit they add this to the list of possible asteroids. By definition for full confirmation, we need to catch the asteroid in another appearance (at least one orbital period later) to verify the identity. The MPC will try and match any detection with something in their archive whenever they get something new. I expect some of the ones we sent in to be surprise follow ups of asteroids that have been seen once.

    The images sub sections of a 4k x 4k image with overlap between the smaller images to catch fast movers. The sections are chopped at the pixel level - which is why there is a one pixel offset that's hard to avoid.

    Much of the image variability is intrinsic to the actual data taken at the telescope. This is real data and it can be rough. That's absolutely the case. As for doing the vector computations - well, that's actually why this work is being crowd sourced. People have amazing ability to see the vector of motion - and identify possible targets that the machines just flat out can't locate. From there we'll turn it into something the MPC can deal with.

    Posted

  • hightower73 by hightower73

    so please ignore my ignorance here, but if the MPC were to out source a program that could search thier databases of asteroids and could compute thier paths, wouldnt it be a case of simply pointing a telescope if the right time and place and check its there???

    i know telescope time is expensive, but surely with people doing this kind of work, surely a program would elminate 75% of thier work?

    Posted

  • grums by grums

    Thanks for the info Dr A. I am surprised that some of the image sets (acquired contemporaneously) don't have the same asteroids in them so that it would not be necessary to wait for the next complete orbit to compute a reasonably accurate orbit. In any case, how can the 4 closely spaced dots give a sufficiently precise vector to be sure of the likely orbital position (in space and time) a year or so later? Surely the MPC would need another point at some greater time and distance than a few minutes and a few arc-seconds? If the system is working this way (awaiting, at least, the next asteroid orbit) it would take several years to verify an asteroid wouldn't it?

    The corrollary to that would be that, although the desired area of the sky is covered in one period of time, I suppose some asteroids will get by-passed because they could be moving so as to be missed by the telescopes??

    I thought that the jitter maybe due to the resolution limit, but having looked at the images for quite a while now, I note that the jitter can be more than 1 pixel and also is usually systematic over each image; i.e. all the stars jitter in the same direction so could be largely corrected by a small overall shift. I can think of a few reasons for this; but it would not be too hard to post process by picking a few stars (based on a suitable algorithm) and correcting the image to get an average alignment of each frame. The algorithm needs to pick dimmer stars to do this as bright ones with a large resulting image (or other objects) would not be so precise. Actually, as this is not a trivial issue given the image variability, a method to allow users to manually re-align images (1 or 2 pixels in any direction) may be easier. However, I can imagine any of this work may be thought of as not cost-effective compared with effort to get a space based observation/detection system in place.

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    @hightower73 - Much of the MPC's work is semi public (the algorithms are published and there are other resources) but they are charged with being the arbiters by the International Astronomical Union. Right now they're going through some transitions - but we'll see how things play out. (The hold up is us much moreso than the MPC)

    @grums - With 4 images you get a set of possible orbits. The MPC has some amazing abilities to connect things that were seen many years ago with something you see today. But you're absolutely correct - for an asteroid to be official we have to see it twice.

    For computing the locations of the asteroids - we determine the center of the clicks and then work out which object is being clicked on. We compute the centroid of the object in sky coordinates (each image has a conversion from pixel to RA/Dec) So the jitter in the images doesn't really change our ability to get the sky location of the asteroid. Whenever sub-pixel shifting is done there is a significant risk of changing the data. The Hubble Space Telescope has quite a large staff dedicated to making their images work with sub-pixel shifts.

    The Catalina Telescope just isn't quite Hubble. It's just not as stable or well calibrated for us to dive into doing that level of shifting - in part because that is the sort of error that humans are fantastic at ignoring.

    Cheers!

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    Since I said I would update: The MPC wants us to work on formatting a bit more - since we're using Catalina data - but AsteroidZoo isn't quite the same as the observatory we have a couple extra things we need to do. We've done those things and are waiting on the next iteration.

    Thank you for your time:

    Posted

  • Barbalbero by Barbalbero in response to Dr.Asteroid's comment.

    Thanks for the update.
    Just one curiosity if possible and if it does not take too much time: which images have been sent up to now to MPC? If I remember well 35 sets of images have been considered up to now

    Posted

  • hightower73 by hightower73

    Formatting? OK in what way do they mean formatting? I guess they mean looking through images? What are the other couple of things they wanted us to do?

    What is the next iteration the asteroidzoo team are hoping to do?!

    Posted

  • Emili_Sancha by Emili_Sancha in response to Barbalbero's comment.

    Where do you get this information? -> "35 sets of images sent to MPC"

    Regards
    Emili

    Posted

  • Barbalbero by Barbalbero in response to Emili Sancha's comment.

    From a previous comment of Administrators

    Posted

  • planetaryscience by planetaryscience

    A search of potential asteroids in the area yielded nothing; the asteroid appears to be magnitude 18-18.5.

    Posted

  • planetaryscience by planetaryscience

    PS I can't find the asteroid zoo link for it, but the original images are at:

    http://asteroidzoo.s3.amazonaws.com/CSS/703/2012/12Apr01/azoo/01_12APR01_N21022_0001-26-scaled.png

    http://asteroidzoo.s3.amazonaws.com/CSS/703/2012/12Apr01/azoo/01_12APR01_N21022_0002-26-scaled.png

    http://asteroidzoo.s3.amazonaws.com/CSS/703/2012/12Apr01/azoo/01_12APR01_N21022_0003-26-scaled.png

    http://asteroidzoo.s3.amazonaws.com/CSS/703/2012/12Apr01/azoo/01_12APR01_N21022_0004-26-scaled.png

    Posted

  • Barbalbero by Barbalbero in response to planetaryscience's comment.

    Is it the result obtained by the MPC?

    Posted

  • planetaryscience by planetaryscience in response to Barbalbero's comment.

    I did a search using this

    Posted

  • BartJSimpson by BartJSimpson

    The asteroid is a known one. It is the main belt asteroid (2909) Hoshi-no-ie with an orbital period of 1915.6794532 days (5.24 years). The asteroid was discovered on May 9, 1983.

    Posted