[Prev][Next][Index][Thread]
(meteorobs) Fireball Network
Ed Majden wrote (about a video fireball network in Vancouver):
>Sandia Laboratories of New Mexico heard of our activities and selected
the University of >Victoria and this network to operate a video
recording system that they have developed. This >device is similar to
the all-sky cameras used during IGY to record Auroras, except that it
uses >a video camera to image the sky in a convex spherical mirror.
This device will record the >fireball on video tape. Again, inexpensive
home type VCRs are used, recording on T-160 >tapes in the extended play
mode providing 8 hours of recording time on a single tape. It >would be
too time consuming to monitor each recording so the public still has to
be used to >report fireball events with the time.
I just read about your fireball network on meteorobs and was wondering
if you could fill us all in on some additional details. How many camera
stations do you have set up and what type of limiting magnitude do they
achieve? Does the video system use an intensifier? What is Sandia's
interest in all this? Sounds like a neat project.
What I thought Ed might be interested in knowing (and anyone else out in
meteorobs land) is that I have gotten the next generation of automated
meteor detection software up and running on a PowerMac 9600 with a Scion
frame grabber board. The system operates in real-time on FULL
resolution video imagery working from pre-recorded videotape or from a
direct camera feed ("live") mode of operation. The previous code I had
only worked at one-half resolution. The development work for this
project was funded by an NSERC grant to Dr Hawkes of Mt. Allison
University, Canada. The software "MeteorScan" was recently delivered to
Dr.Hawkes and has been successful in detecting 80% of meteors visually
spotted by manually scanned pre-recorded tapes. I am currently working
on an upgrade that will hopefully improve on that. This may be useful
for Ed's network in not requiring the public to inform them of a
fireball event, but rather to scan the tapes automatically to find all
meteoric events or possibly work in a live mode of operation.
The software is currently set up to be hosted on the computer/digitizer
system described above but could be easily ported to other systems since
the frame grabber functions have been broken out separately from the
detection algorithms. It would simply be a matter of rewriting the
digitizer control functions for the frame grabber of interest. The
software uses a localized Hough transform algorithm to detect lines in
video data streaming in at 30 frames/second. The software saves a strip
of the detected linear track of the meteor of duration one-half second
and a single full frame for star background registration. It turns out
that the 233Mhz system I am running is plenty fast enough to run the
detection algorithms in real-time and that the main bottleneck is the
PCI bus speed in transferring data from the frame grabber to computer
processor (50 MHz). With a current generation PC with 100 MHz bus
speeds, the overall system performance should be greatly improved. I
have been operating the system in live mode from my rooftop and
typically detect ten meteors overnight in light polluted skies.
Unfortunately, weather has been pretty miserable lately on the east
coast of the US so I haven't gotten many nights in with the system
operating.
Pete Gural
pgural@trg1.saic.com