Asteroid Zoo Talk

"Bad set" update: Checkbox now added to classification interface!

  • DZM by DZM admin

    Hey everyone,

    So, the team and developers have just unrolled a new way to get rid of those pesky bad sets that everyone has been talking about.

    Next time you classify, you'll notice a nifty little checkbox underneath the "Add to Favorites," in the bottom right of the interface. When you find a bad set, just mark it as so! That will flag the image, and after it's flagged enough times, it will be removed.

    This system is brand new, so if you have any bugs or issues with it, or additional questions, please let me know here.

    Hope this helps with the issue. Feedback is very welcome!!

    Posted

  • hightower73 by hightower73

    thats a terrifffiiiiiiiccccccccccccc idea! no there should be no excuses lol thank you dzm

    Posted

  • Xabo by Xabo

    Is all bad sets will be eliminated when marked ? I found a rather large asteroid in a bad set... so some of them could contain valuable information.

    Posted

  • stonepenny by stonepenny

    define a 'bad set', please? I've found known asteroids in 'bad sets'....
    Ok, a good set - 4 clear photos.
    An iffy set, 3 clear, one fuzzy/blank/white out, or a part photo?
    Poor - 2 photos from different sets, together as one set. I've found potential asteroids, ie a spot that moves over 2 photos in these.
    (where 2 are from the last set, I've already done that one, so ignore them) or two that are fuzzy/blank/white out, or a part photo.
    Bad, no chance of seeing anything over 2 or more photos. (Blank/black/fuzzy) so no way to track an asteroid.

    The above are just some thoughts... some guidelines from you would be much appreciated, please.

    Posted

  • DZM by DZM admin

    I'll alert Dr. Asteroid and see if he can answer your questions! 😃

    Posted

  • Dr.Asteroid by Dr.Asteroid scientist, admin

    So, for what we typically consider a "bad set" I'd say would be where 2 or more frames are unusable.

    If one of them is off - that's "sort of" okay - since we need three points to define an asteroid (mandated by the minorplanet center) but if two are really unusable - tag it and let's get rid of it.

    Posted

  • grums by grums

    Seems a reasonable criterion. But good luck finding the orbit of an asteroid with 3 rough points from a 256x256 array on a picture taken many years ago. I think many people here would like to know exactly what are your procedures for confirming asteroids. I guess it would be quite time consuming and probably require getting particular asteroids in more than one set to get a better fix.

    It would also interesting to get a feel for the cost (in terms of time) effectiveness of searching methodologies. I note that although some asteroids are reasonably obvious, most "possibles" are really at the level of background noise and could be the effects of occluding clouds or simply due to random fluctuations in the background that happen to produce points in an equally spaced straight line. It may be better to spend less time and cover more sets for example.

    What is clear, when you search for patterns in noise, is that having more points increases the chance of being correct as opposed to finding a random noise sequence. This would be true for automated as well as human searching techniques so it would be worth, in future, considering more frames (say 5 or 6) at the expense of telescope time. It would also, of course, be good to get rid of a lot of atmospheric distortion by using a satellite based telescope or a telescope with artificial guide star correction but I know this is expensive. With the newer 1.4Gpixel CCD arrays and telescopes with these features in place, and specifically looking for NEO asteroids, what are the plans for this program?

    Even with these 256x256 frames much could be done with post processing algorithms to reduce clutter and make both automated and manual detection simpler and faster.

    I guess you guys have thought about all these issues but it would be nice for us to know a bit more.

    Posted

  • grums by grums

    Oh dear, sorry: I have loads of questions as I am ignorant of how all this work is progressed and am very interested to understand more. Here are a few...

    1. I see that the CSS telescopes are said to be equipped with cryo-cooled CCD arrays of 4k x 4k (and the one in Australia with much higher resolution). Why are the images we see 256x256? I can think of many possible reasons but it would be good to understand which are correct.

    2. As the images we see seem to not be looked into for a considerable time (an example was a set I saw with an obvious asteroid that was pointed out 6 months ago for example), how do you verify them? Do you get a rough track off the 4 closely spaced points and then find a contemporary set which that track would intersect then see if the asteroid appears in the right place? And, by doing so, also get enough precision to be able to get an orbit determined? I can see there may be no urgency as it will be some years before any NEO asteroids would come close to the earth again.

    3. Are there any future plans to look into using a CMOS sensor array rather than CCD? They are getting better in terms of sensitivity and resolution and have the advantage of not being so affected by the high dynamic range that a nearby bright star can inflict on an image. They are also becoming standard technology on modern, mass-produced processes.

    4. Would use of a guide star (artificial if necessary though maybe not needed) enable better rejection of atmospheric distortion? The wavefront correction technology is quite expensive but I note that this can be done without the use of mirrors etc. with CCDs that can also shift orthogonally and possibly with just post-processing with CMOS detectors (not sure about this). I appreciate this is future technology.

    5. I thought more post processing could make visual detection easier but I have not found a way that could be applied generally to all image sets. It is easy to see some asteroids but others can not be distinguished from fluctuations in the background (from occluding clouds for example). Is it better to look through lots of sets for the more obvious ones or spend ages poring over a few sets trying to pick out dim asteroid tracks from numerous possible ones that may well be random events? Without being able to weigh the advantages and disadvantages without adequate knowledge means that it would be better to have guidance from you guys based on some statistical analysis.

    6. Further to 4 and 5 above, removing jitter would help significantly. As there are usually plenty of small star images present could these be used as Guide Stars to post process the images to remove jitter globally and locally? It may not be worth the effort though.

    7. Why only 4 images per set? Whether visually or automatically trying to detect the asteroids, the key factor that distinguishes a moving body is the point images being in a straight line and at equal spacings. Given that a major problem is distinguishing this from random background fluctuations, having a larger number of points (more than 4) has a dramatic effect on reducing false hits. Again this is a future consideration which may or not be a good trade off with telescope time and with improved imaging.

    Best regards and have a great New Year

    Graham

    Posted

  • peterbees by peterbees

    OK Grums, though having no inside information I will attempt to answer your questions (in the absence of any other response).
    You are of course correct that the project would be more satisfying and probably more productive if users were provided with more and better images. However, consider the following:

    1. This is not a premier, highly funded and resourced research project. To cover the sky with state-of-the-art resolution images dedicated to asteroid hunting would be hugely expensive and intrusive on other projects. Hence the researchers work with what is available, together with the free services of enthusiasts like you and me.

    2. The images are not as bad as they may look (which is a bit like saying that Wagner's music is not as bad as it sounds...). The pixel count may be only 256x256 but I suspect that these are subsets of larger images. In any case the sky area is very small. Just having access to such data on a home computer would have been unthinkable 15 years ago.

    3. More importantly consider the stated aims of the project. There are some half million known asteroids, identified and categorised at great expense and effort since 1801. How many new objects can the project add using the scrappy image set we are presented with? I have no idea, but at a guess a 10-20% addition to the current tally might be possible. Even 1% would be 5000 new objects to track and document. That's a major advance, which needs lots of eyes to sift through the current data rather than new sensors or orbiting platforms to achieve.

    4. The researchers have stated clearly that one of their aims is to develop algorithms to automate image enhancement and object identification - and that prior attempts to do so have not been able to match the skills of an experienced human observer. From this point of view the project participants are simply providing a baseline data set which will be used to evaluate the ability of updated computer programs. The identification of a potential asteroid by a single user is therefore important even if it has been tagged many times before.

    5. Whether or not 'new' objects are followed up to the point of confirmation and registration with the appropriate authorities (the IAU?), the process will take a loooonnng time. Don't expect naming rights in the foreseeable future.

    Along with others I have a number of misgivings about the design and implementation of the Asteroid Zoo, however it seems to be a 'fit-for-purpose' attempt to do significant research on the cheap. Also, like others, I get addicted to finding those little moving blobs.

    Hope this helps.

    Posted

  • nicro46 by nicro46

    I agree Peterbees, I am convinced that we are trying to get as much information as possible from one mole of images is not very recent and not great quality, however, is seil our work can be helpful we are available. As for a sensitive topic, if and when the IAU find 'between those we reported new asteroids and must give a name, it would be really great that they could also use the nikname of first zootie who reported it!

    Posted

  • grums by grums

    Thanks for providing some comments Peter. Yes, I do understand that this is not hugely commercially funded and I really like the concept of crowd sourcing operations that, at least at present, can do better than than software. As someone new to this, and reasonably enthusiastic having just hit 1000 image sets, I also find it quite fun looking for and finding a "possible" asteroid in the scintillating porridge in some of the image sets. It is just mainly that I am curious as to how the whole system works and I, and I think others, may be more enthusiastic if there was some more information on how it all works and also see what the future holds for this and other programs.

    I have spent 40 years working in semiconductor design and am very interested to know more about this side of the technology for example. Cryo-cooled CCDs can be very impressive in terms of having a low dark current; often down to a few electrons during an exposure. But they have other disadvantages as can be seen in some of the bad datasets. It is really just general curiosity and the quickest way to understand things is usually to just ask. Even just being pointed at scientific papers or reports would be good. I would like to see what the future plans are and would be happy to see us all made redundant by technology improvements 😃

    I don't care so much about having my name associated with an asteroid, nicro. Just think if it was your asteroid that was about to wipe out life on planet earth 😃

    Posted

  • nicro46 by nicro46

    The idea would be to use the "nikname", not the real name that is normally used to famous people, and only once. A designation would still almost anonymous, instead of an alphanumeric code. On how we should hear other suggestions, but I fear that this is just an academic discussion. Regards grums.

    Posted