Author Topic: **HOW AND WHAT TO CLASSIFY**  (Read 18454 times)

jules

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1416
    • View Profile
**HOW AND WHAT TO CLASSIFY**
« on: February 23, 2010, 11:22:30 pm »
OK. What I'd do (and please don't take this as definitive, just a suggestion for discussion) is to classify events as follows;

1. frames with missing blocks or combined images (ie frames with something obviously wrong in them or bad data) as 'something else'
2. particle strikes as any streaks, faint or otherwise, that are not vertical and not coming from the direction of the Sun.
3. any rings, flares or arcs that are associated with (or occur at the same time as) bright planets in the field of view (or just entering or exiting) or that do not change position much over sevaral frames  (the last example being a particularly good one of this) as 'optical effects'.
4. 'Comets' as any object that crosses the field of view in any direction with, or developing, a tail. Some comets falling into the Sun appear as very fast moving bright streaks on the Sunward side of the image, others have well defined tails (pointing away from the Sun) and others initially appear as small dots moving across the starfield that develop tails as they approach the Sun. We can provide examples of all these types.

What we *don't* need tagging (simply because they occur almost all the time and are well-understood) are the vertical lines (bright or otherwise) that are associated with bright objects or stars in the field of view. This is just excess signal that has been allowed to 'bleed' into pixels above and below the bright object rather than saturate the whole image (see my waffle analogy elsewhere in this forum!). equally, occurrences of the Milky Way are pretty obvious and predictable so we don't really need you to be them (although seeing anything else when the Milky Way is in the in the field of view is certainly a challenge!

Thanks again for your time and efforts so far,

Chris.
« Last Edit: February 25, 2010, 09:59:23 pm by jules »

jaynna

  • Newbie
  • *
  • Posts: 1
    • View Profile
Re: HOW AND WHAT TO CLASSIFY
« Reply #1 on: February 25, 2010, 09:34:38 am »
I was wondering how I should mark the timestamp for an optical effect which lasts for the whole clip. In this clip here http://solarstormwatch.com/favourites/165 there is an arc in the beginning of the clip (look to about 1.5 cm for the right edge) and it's there through the whole clip, just more or less visible.

aurel

  • Newbie
  • *
  • Posts: 6
    • View Profile
Re: HOW AND WHAT TO CLASSIFY
« Reply #2 on: February 25, 2010, 11:43:27 am »
Chris said in one of the intro videos that we should mark the beginning and the end end of such an effect.
So you'd place a mark right at 0:00 and one at the end of the video. Placing one at the last frame is a bit tricky because when the video finishes it resets to the start, but if you pause it short before and then use the "next frame" button it works good enough.

boojum_uk

  • Newbie
  • *
  • Posts: 2
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #3 on: February 26, 2010, 01:38:26 pm »
What do we do about the flashes which comprise a black bar with a lighter band extending from them to the top of the frame?

ElisabethB

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 689
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #4 on: February 26, 2010, 01:51:30 pm »
hi boojum_uk
I consider these 'missing blocks' and mark them as 'something else' (see point 1 in the first post).
Welcome to Solar Stormwatch !  :D

boojum_uk

  • Newbie
  • *
  • Posts: 2
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #5 on: February 26, 2010, 03:56:50 pm »
Yeh, I guessed it was something like that and have been classifying them as 'something else'

2 comets so far - woohoo!   ;D

StruckByRain

  • Newbie
  • *
  • Posts: 2
  • Eh.. You'll learn more about me when you talk to me. :P
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #6 on: February 27, 2010, 09:00:02 pm »
Lol. I don't think I have seen any comets yet. Lol *lucky^* XP

yankeegal01

  • Newbie
  • *
  • Posts: 10
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #7 on: February 28, 2010, 04:36:17 pm »
How about when the image "jumps".  I don't mean missing images, or the image stops, well, the best way to explain is to show I guess.  In the following example around 14.30 or so (give or take a nano sec).

http://solarstormwatch.com/favourites/434

srcrothers

  • Science Team
  • Sr. Member
  • *****
  • Posts: 329
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #8 on: February 28, 2010, 08:31:28 pm »
This looks like an example of the HI 1 B pointing being disturbed and upsetting the background subtraction that has been discussed before on the forum.  Since we do a fit to the stars in the field of view in order to solve the problem we already have good information on what happens when and so it is not necessary to report these

thanks

Steve

lpspieler

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 329
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #9 on: March 01, 2010, 12:00:47 am »
Through various posts, I've noticed that there are conflicting instructions on how to mark persistent features in what's that. The written instructions tell us to mark the single frame where a feature is visible best. But the instruction movie by Chris tells us to mark the beginning and the end.

I personally prefer the second alternative, but a definitive decision should probably be taken.

Quialiss

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 494
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #10 on: March 01, 2010, 01:12:17 am »
Well, some features, particle strikes, missing data, bright flashes at the edge of the frame.. are only single frame events, or have a definite 'most visible' frame to them so it makes sense to mark them just once.  It's the big rings and arcs that are hard to do that way, because they generally show up for a good part of a video.  It would be awesome if we could mark blocks of time as an event, but for now start/end markers will have to do.

skelband

  • Newbie
  • *
  • Posts: 1
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #11 on: March 09, 2010, 09:40:33 pm »
When spotting features rather than looking for storms, should you mark storms as "Something Else" or just ignore them?

ChrisScott

  • Science Team
  • Sr. Member
  • *****
  • Posts: 332
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #12 on: March 09, 2010, 09:51:45 pm »
Just ignore them in the 'whats that?' activity. We'll have the info about them from the 'spot' activity.

Thanks so much for your time!

Chris.

gimre

  • Newbie
  • *
  • Posts: 5
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #13 on: March 14, 2010, 05:32:28 pm »
Hi

I'm new here, looking at these only for 2 days. Is it just me, or does the STEREO B contain far less particle strikes than STEREO A ?

Does everything that's a "fast line on a frame" qualify as a particle strike, or just those that look more like a "splash" ? I'm noticing 3-4 (at times even more) of these fast moving (sometimes really blurry) lines going through in one frame. I really don't want to mark false positives.

And one other thing: it seems to me that the "Incoming!" videos repeat themselves. I'm not 100% sure, but the last 3 times I think I saw the same exact video.

All this seems like a really great idea, I'm having fun trying to figure things out :)

Quialiss

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 494
    • View Profile
Re: **HOW AND WHAT TO CLASSIFY**
« Reply #14 on: March 14, 2010, 06:08:47 pm »
Hi gimre,

I haven't noticed any differences in frequency of particle strikes between A and B, but if there is, you can bet it will show up when they look at all the data, and heads will be scratched.   ;)  Your 'fast lines' sound like particle strikes to me, they're often dim, fuzzy, and otherwise hard to spot.  You can see some examples here.

The incoming video updates very slowly... each frame is 40 minutes apart, and sometimes there are gaps in the data.  So the last few frames will always be new when you watch it, but the majority of the video will be the same.

I'm glad you're enjoying the project!
Quia