Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!zombie.ncsc.mil!news.duke.edu!godot.cc.duq.edu!newsfeed.pitt.edu!gatech!howland.reston.ans.net!math.ohio-state.edu!news.acns.nwu.edu!news.eecs.nwu.edu!k9ape
From: k9ape@eecs.nwu.edu (Sheldon L. Epstein)
Subject: Re: VIDEO CAMERA WITH VARIABLE NO. OF FRAMES PER SECOND
Message-ID: <D5E5o8.KpI@eecs.nwu.edu>
Sender: usenet@eecs.nwu.edu
Organization: EECS Department, Northwestern University
References: <3jm6oq$2gr@dingo.cc.uq.oz.au> <D59BDK.577@eecs.nwu.edu> <DBwOleM7KpvO078yn@abc.se>
Date: Mon, 13 Mar 1995 18:21:42 GMT
Lines: 74

In article <DBwOleM7KpvO078yn@abc.se>, Mattias Ernelli <m8939@abc.se> wrote:
>In article <D59BDK.577@eecs.nwu.edu>,
>k9ape@eecs.nwu.edu (Sheldon L. Epstein) wrote:
> 
>[Stuff deleted]
>
>> Next, the bad news,  These cameras only operate in the DONPISHA
>> mode at shutter speeds of 1/2000 or less, are limited to one NTSC
>> or PAL field every 1/30 or 1/25 second and operate in the field
>> mode.
>> 
>> Unless the centrifuge is turning slowly, these cameras won't produce
>> sharp images.  My suggestion is to use these cameras at higher shutter
>> speeds and not try to get one frame per revolution.  One camera,
>> a PULNIX TM-9700, is capable of delivering full NTSC frames at
>> shutter speeds of 1/16,000; however, performance is noisy at that
>> speed - partly due to light source flicker.
>> 
>
>[Stuff deleted]
>
>I have been working with the SONY XC-77RR in donpisha shutter mode in a
>project where it was important to capture an image within a time
>interval less than 20ms, an asynchronous camera was the only solution in
>this case. In the situation which the orginal poster was in, I think
>that one solution would be to synchronize the revolving centrifuge with
>the camera. For a revolving of say 100 RPS (Turns per second) the
>centrifuge would be synchronized to twice the framerate of the camera
>(PAL 50Hz). The SONY camera has this bad limitation of only 1/2000'th a
>second shutter speed when running in DONPISHA mode, but in normal mode
>very short integration times could be used (up to 1/1.000.000 sec i
>remember).
>
>If the speed of the centrifuge can be controlled, which it probaly can
>be, then an external PLL circuit which locks on a multiple of the
>vertical sync could be used to controll the speed of the centrifuge.
>
>Another solution, if above is not possible, would be to use the SONY
>camera (in donpisha mode) set for say 1/2000 a sec of integration time
>and use a flashlight with controllable duration in the 1/1000 -
>1/100000s range to controll the exposure. Since the donpisha camera has
>to be externally trigged, the same trigger pulse fires the
>flashlight/strobe. But that solution only works if the centrifuge speed
>is less than 50/60s RPS. Anyway, 50 RPS is 3000 RPM which is quite a
>high rotationspeed for a centrufuge i belive.
>
>Hm, when I am thinking, the centrifuge speed is not important actually
>when using donpisha mode, only the exposuretime (to prevent motion
>blur), only if analysis is to be performed at each turn, in which the
>camera has to be fully synchronized with the centrifuge. But at a rate
>of 50 fields a second, I cant think of any PC-Framegrabber based
>solution that can handle that analysis speed.
>
Hello Mattias,

Now you have me thinking about the centrifuge problem again - and I 
think we can do it.  The trick is to use the SONY XC-77RR camera in
the Donpisha mode with a strobe and couple the camera to a brand
new (we just received serial number 2) EPIX COC40-2 Imaging Co-Processor
Board hooked to an EPIX Model 12 Imaging Board.  The COC40-2 Board
has twin TMS32C40 50 MHz DSPs that can process independently or
in parallel mode.  For example, one 'C40 could be processing
while the other is capturing.

I saw an EPIX demo showing the use of the card for real-time target
tracking and it was quite impressive.  The board set was capturing
30 frames (60 fields) per second, tracking the movements of the
target and keeping the cross-hairs on the target.  If the centrifuge
problem is one of comparing the current frame to a past frame and
analyzing the difference, then the EPIX COC40-2 Co-Processor board
with the SONY XC-77RR Camera is my recommendation.

			Shel Epstein, k9ape@eecs.nwu.edu

