[Prev][Next][Index][Thread]

Re: (meteorobs) Re: WGN video paper

Here are a few comments arising from a conversion I had today with
Geoff Grayer, and following a quick browse of the article.

I agree with the authors that the video technique has great potential.
Indeed I've long wanted to operate one on a regular basis for some
while, having been motivated since the late '80s by the challenge of
automatic meteor detection and reduction, and well as the potential
for observing faint meteors and investigating radiant structure and
minor showers.  This why I'm a telescopic observer, but I could easily
switch allegiance to video once similar science was achievable, as
more-accurate results are possible because the velocity is measurable.

My goal was to apply some of the techniques I've used for galaxies,
some bespoke image-processing algorithms and recognition tools like
neural nets.  As Sirko says ideas are one thing and working algorithms
are another.  Without real data in digital form it's not been possible
to test these ideas and methods.  That may seem odd since if I tell
you my colleague, Geoff Grayer, and I have had a camera since autumn
1993, and a second in 1994.  Our efforts have been frustrated by
hardware problems.  Most notably we were trying to find affordable
frame grabbers.  We thought we had---they were affordable---but didn't
seem to work, as we could only get 2-bits of information written to
the output file, and an episode of "Star Trek" playing on my PC.  When
we went back to the company, we learnt that they had gone bust, and
our efforts to obtain a circuit diagram or a refund have proved to be
a waste of time.  Our intensifiers were cheap (US$150-350) 2nd
generation MCPs (these were available at least two years before the
time implied in the article).  They have less noise than say MOVIE,
but do suffer from 25% distortion (mostly in the outer third), and
that is a concern for any standard technique, and cosmetic defects.
Taking out the distortion using star fields is quite feasible, but
will need some hefty processing power to remove from an automatic
system.  It's important to correct for distortion if say your
automation is looking for objects moving in a straight(ish) line.  Any
resampling does, however, lose information.  The mark II camera (my
one) showed strong vignetting and Geoff has had it `in the shop' for
about a year playing with lenses.  More recently, the Newvicon camera
(more-sensitive than an ordinary vidicon but it's not digital like a
CCD) has blown.  We bought another voltage controller, but that's not
fixed it.  It's also quite bulky compared with a CCD and needs mains.
There were numerous other things I won't bore you with, but you get
the idea.

All-in-all it's been educational, but ultimately frustrating.  Now the
reason why I've spouted on about this, is not to discourage new
observers, but to indicate that video does require a strong commitment
both financially and in time (my meteor efforts have been mostly
directed to the IMO Telescopic Commission), and more importantly to
support a point made in the article.  Yes IMO needs a Video
Commission. That's been my opinion for some years.  The Commision
would collate information about lenses, image intensifiers, frame
grabbers, CCDs, time-tagging, construction, software, tips on getting
the kit through customs etc., and hence enable new video observers to
get operational systems at affordable cost in a reasonable time.  It
can also co-ordinate observing campaigns for such things as
double-station observations, video-telescopic comparisons to determine
the errors and speed scale of the latter.  It would be a vehicle for
communicating ideas and setting up standards.  It could offer advice
about techniques, e.g. should you track with the sky or leave the
camera fixed in altitude and azimuth.  Are you volunteering Sirko?

Given the above frustrations, I'm prepared to start afresh and invest
in a CCD camera, better image intensifier, and functioning frame
grabber.  So any tips on which ones I should buy or at least
investigate.  I'm also thinking seriously about buying a modern PC for
both professional data reduction, and meteors.  The one I have at home
is a lowly 386/33 with 40Mb of disc space courtesy of work.

Re. recording of spectra, I know that you were trying to cut down on
references, but I think the outstanding work of Henry Soper should get
a mention.


Now to Sirko's reply to Peter.

> > The goal being to avoid storage on tape and
> > work directly with the video signal (lower noise, shorter turnaround time).
> 
> I don't know whether this really is the goal, this is something to be
> dicussed further. I see here two ways: Either all data is digitally stored
> on a PC (or similar device) and inspected afterwards, or you let your
> meteor detection system run continously during the observation and only
> store the frames that possibly contain a meteor.
> Solution 2 (I guess that's what you are heading for) has the big
> advantage, that the amount of data to be stored is only a fraction of
> solution 1. However, I still favour the first method. If the automatic
> system fails for any reason, the data is definitely gone. Not so in the
> first case...

I think you want both.  The raw tape should be an archive of the data
and should be retained.  If improved techniques come along or some
other aspect of the data are to be measured, you can still do it.
Also there may be competing data reduction techniques and we may wish
to compare them.  If the data can be stored digitally so much the
better.  The way things are going with computer technology who knows
what will be possible in a decade say.  The mass market will surely
make high-density writeable CDs cheap.  I'd be reluctant to have
digitally compressed data created with a lossy algorithm.

We may not be able to produce a fully automatic system straightaway
because the processing power is not yet available, so in the first
instance I'd like a semi-automatic system that locates the meteors.
You then just have to transfer the frames that span the meteor event
for analysis.

On the processing power front, Geoff Grayer was keen to do some
pre-processing like averaging and differencing frames in hardware, and
perhaps if an algorithm is devised for automatic detection, this too
might be done in bespoke hardware.  As he's produced similar equipment
for experiments at CERN, he could well be right.  I'm only a software
man!

> > With a bank of imagers this may become necessary or you may get buried in
> > tapes.   What about what the others out there are doing? Is there anyone
> > else seriously looking into this?
> 
> I completely agree: The number of tapes is a serious problem, as long as
> we are in the development stage. I hope that we will have it solved once
> automatic meteor detection is available.

As I said above the VHS tape could be replaced by a CD-ROM sized disc
containing a few hours of digital TV when the technology becomes
available.  I read about it a year or two ago in "IEEE Spectrum".
Ideally you'd like more than one copy at different locations.

> > Another item is that of the use of video systems should also include just
> > basic monitoring of activity levels on a night to night basis.  I just read
> > the article in this months WGN on the survey of meteor observers.  Seems
> > there are big gaps in observational periods for the typical observer.  I
> > know due to work, family, and the need for sleep on occasion, I tend to
> > only observe clear weekends (fri, sat) or within a few days of major
> > showers.  Video systems help fill the gaps when fewer or no observers are
> > out on a watch!

Yes this is most desirable.  In the 1970s in the UK there was a
fireball network, where fish-eye cameras monitored the sky each clear
night.  Even armchair astronomers were enthusiastic participants as
they didn't have to sit out in the cold.  The telescope makers joined
in as they built automatic systems that would detect if the sky was
clear, begin observing, close down at a certain time or when it
clouded over.  Video does offer scope to those who like their creature
comforts, and might attract these types again.  So perhaps when we
have our act together, we should publicise it widely, and not just in
the meteor community.

> This is what we address as the third key project, the determination of
> flux densities.
> [snip]
> With video systems we are able to monitor the activity every night we
> want. However, it would make no sense to report this activity from video
> observations in terms of ZHR to compare it with human observers. The
> ZHR is a construction that describes at it best what a human observer
> sees. From that we later calculate flux densities (i.e. the real number of
> particles hitting Earth at a given time and area), which is not so easy
> because we have to assume certain things about our ZHR. So it makes much
> more sense to derive the flux directly from the video records, whose
> detection capabilities are much easier to determine.

Agreed.  No need to invent an equivalent for ZHR.  Just go to the flux
densities directly.

Maybe more later once I've actually read the article fully and carefully.

Malcolm
Follow-Ups: References: