PDA

View Full Version : 3D Tracking Solutions



Steve Boyd
03-11-1996, 01:50 AM
Thanks to all who responded to my email regarding the fluorescent marker
tracking problem I had. A copy of my original posting is followed by some
of the suggested solutions to my problem.

To quickly summarize, I will try a high frequency electronic ballast that
fires the UV light at above 25 kHz which should resolve my 'beat' problem.
These ballasts are available at any commercial lighting store. I am also
looking into using reflective paint rather than the fluorescent orange paint
as a backup plan if the electronic ballast doesn't do the trick. If anyone out
there has had experience using reflective paint for motion analysis, their
comments
would be appreciated.

Thanks again for all the responses, any additions to the following list of
responses is welcome!

************************************************** ****************************
My original email:
************************************************** ****************************

Dear colleagues:

Due to the small size of the markers I would like to track to obtain 3D motion
analysis data, I have chosen to use fluorescent orange markers rather than the
traditional reflective tape (the tape is too difficult wrap on the small
markers, ~1.5 mm diameter, and reflective paint is hard to find). However,
using the fluorescent markers, I have encountered a problem that I hope someone
else may have experienced. The images of my fluorescent markers are fading from
visible to non-visable during tracking.

The problem is that I collect my video data at a certain fixed frequency (60,
120, 180, or 240 Hz). The fluorescent UV light source used to illuminate my
orange markers gets its A.C. power source from a wall socket which has a
non-constant frequency of close to 60 Hz. The small and fluctuating difference
between my video data collection frequency, and the UV light power source
frequency causes 'beats' in the intensity of my marker video image.

It should also be noted that that video cameras normally operate using red LEDs
that are synchronized with the camera collection frequency. Therefore, I have
access to a 'synch' signal and a 12 volt D.C. power supply from the camera to
operate the fluorescent light source. Unfortunately, I don't think 12 V D.C.
fluorescent lamps are available.

Some possible solutions I have come up with are as follows, each with their own
problem:

1. Use a constant UV light source (however, low intensity of the UV source
becomes a problem).

2. Synchronize the UV light with the output signal from the camera (I don't
know of an A.C. generator where the frequency can be controlled by an external
signal).

3. Control the frequency of the wall power source to be exactly 60 Hz (I've
been advised this is not easy to do).

If anyone out there has any comments or other possible solutions I would very
much appreciate the input. I'll post the results.

Sincerely,
Steve Boyd

M.Sc. Student at:
Human Performance Laboratory,
University of Calgary,
Alberta, Canada.

************************************************** ****************************
Some responses:
************************************************** ****************************
Steve:

Suggestion:
Use a high frequency inverter to run the UV tubes. This will run at about 1
to 5 KHz. Either get parts and build one ( difficult unless you can find a
kit... there are several here in Aust. but don't know about the USA ) or
try a 12 Volt fluorescent of the type used in recreational automobiles or
caravans ( mobile homes etc) or boats and replace the tube with a UV tube.
Most of these use small 10 to 13W tubes so you would need to use a suitable
UV tube.

Another suggestion: The video camera may be able to be synchronised with
mains frequency. However, Mains is not a stable frequency reference,
certainly not as stable as the crystal locked oscillator in most cameras.
This could be a problem in timing, velocity measurements.

John Yelland

.................................................. .................
Technical Services Unit, Faculty of Health Sciences, Carlton Campus,
Latrobe University, Melbourne, Australia.
Tel. (03)285 5332 | Fax. (03) 285 5111 | Locked Bag 12
John Yelland : j.yelland@latrobe.edu.au | Carlton South P.O.
John Horan : j.horan@latrobe.edu.au | Victoria 3053
.................................................. ..................

************************************************** ****************************
************************************************** ****************************

Steve just a quick thought for what it's worth. Most labs will have three
phase power with a neutral line available. That means you have available
three normal power outlets 120 degrees apart. I think the florescence
lights will be illuminated for more than 60 degrees twice during each cycle
so if you have three lights beside each other the output should be fairly
constant. I haven't tried it and would like to know how you get along.


regards

Graeme Shea
Snail Mail 82 Baillie St Horsham Vic Australia
Email spod@netconnect.com.au
Phone +61 53 826406

************************************************** ****************************
************************************************** ****************************

Dear Steve;

I saw your post concerning synchronizing UV illumination and video
cameras and thought I could offer some assistance.

1: Advance Transformer makes the "Advance Mark V Integrated Circuit"
electronic ballast. It is available in 120V (90V-145V) or 277V
(200V-320V) versions and will run on AC or DC. Magnetek makes a
similar ballast, the "Magnatek Triad." Both are available through
the Grainger catalog if you can't find them locally. If you have
access to a filtered 120VDC supply, these ballasts can give you "no
flicker" illumination.

2: If you are only using one camera, it can be synched to line. This
is not a great idea, as line can be dirty, and at the higher speeds,
you will still have light and dark fields, as the phosphors have
fairly short lag. If you _really_ want to pursue this, contact me
for details.

3: Use a xenon strobe (rich in UV) for illumination and synch it to
the camera. This could work if you have a strobe that will work at
240 pps. If you have to buy one, though, it may be too expensive.
Again; contact me if you need interface details.

I think your best choice might be electronic ballasts, as this will
give you flexibility with regard to choice of lamps and illumination
available. You can easily replace the ballast in existing lamps at
low cost. I use both the Advance and Magnetek models, and prefer
the Magnetek ballast since it will start lamps at lower temperatures
and produces less radio interference.

************************************************** ****************************
************************************************** ****************************

Hi Steve:
Here's a few atrocious suggestions:

> 1. Use a constant UV light source (however, low intensity of the UV source
> becomes a problem).
...I think you're right about the intensity, you need a 'flash' of light
synchronized with the LED's.

> 2. Synchronize the UV light with the output signal from the camera (I don't
> know of an A.C. generator where the frequency can be controlled by an
external
> signal).
Get hold of an old tube audio power amp, use the sync signal from the LED
drivers as an input. Connect the output to the 12v secondary of a 110vac-12vac
power transformer (20 watts or more). Connect the 110V winding to your UV lamp.
This might work! Make sure that whoever does this, knows something about
electronics and the relevant safety codes..... If your UV lights have filaments
on them, disconnect them, and try for a higher voltage transformer - why waste
power on the filaments. They work more efficiently (and faster) but require a
lot more voltage. The 8 foot instant-start fluorescent lamps use no filaments,
but require almost 1000v to start. Don't try this if you're not comfortable
with playing with high voltage stuff.
Don't use a transistorized amplifier with the above experiment, it'll
probably blow.



> 3. Control the frequency of the wall power source to be exactly 60 Hz (I've
> been advised this is not easy to do).
...It's probably closer to 60 Hz than your 3D system. Most systems use the TV
vertical rate which is 59.75 Hz, not 60. In addition you'd need to phase lock
the two systems, something that adds a bit of difficulty.


-Paul

************************************************** ****************************
************************************************** ****************************

I would personally stay away from trying to synchronize the powers, unles you
actually went to a UV strobe.

Try this:
1) determine the total power consumtion needed to drive your UV lights. Just
and up all the Wattages.
2) locate a SIGNAL GENERATOR at your school which can produce that much power
plus more. Use it to power the UV lights. (try the science or engineering
depts. if you can't find one which generates enough power, you may need to
amplify it, but you probably won't need to).
3) set the signal generator on 'sine wave' and at a frequency at least an
order of magnitude higher than the collection frequency. keep trying higher
freqs until you reach one that works. (ideally, you want to know how long
the camera is samples for each frame, that is it's duty cycle, and at least
double the inverse of that time period. but that's not always easy to
interpret)

Best of luck,
Terry George

************************************************** ****************************
************************************************** ****************************

UV "quenching" of flourescence is a common problem. I would suggest that you
try retroreflective paint. Although it is only 10% as reflective as films,
it should work well for your application. It will also permit you to to use
visable or IR illumination.

In the US, technical information on 3M Reflective Liquids is sold through
the 3M Commeercial Graphics Group 1-800-328-3908. There is also a local
Calgary number for 3M products:
403-275-7330.

The material you want to ask 3M for is:
Series 7200 Reflective Liquid. Ask for Instruction Bulletin 5.3. This will
describe the application and reflectance.

The Canada office for 3M is in London Ontario: 519-451-2500.

Good Luck,

Bill Hand
AOA/QWest

************************************************** ****************************
************************************************** ****************************

>1. Use a constant UV light source (however, low intensity of the UV source
>becomes a problem).

This is the easiest option. If the intensity is too low, use lots of them.

>2. Synchronize the UV light with the output signal from the camera (I don't
>know of an A.C. generator where the frequency can be controlled by an external
>signal).

There are quite a few designs for inverters that generate mains voltage
from 12V or 24V DC - they crop up periodically in electronics magazines and
you can usually buy kits too. These will explain the circuit in detail, and
you (or a friendly technician) ought to be able to alter the design so that
oscillator can be synchronised to an external clock pulse fairly easily and
safely (that's what opto-isolators are for).

>3. Control the frequency of the wall power source to be exactly 60 Hz (I've
>been advised this is not easy to do).

You'd need to convert it to DC, then chop it up again to produce AC at the
frequency you want. Again, it's fairly easy. If you have a good electronics
person, this will be the cheapest option, but you need to be careful since
you're dealing with high voltages and mistakes will fry things :->

Good luck
Bill

----
Dr. Bill Sellers, Email: bill.sellers@ed.ac.uk
Department of Anatomy, Tel: UK (0)131-650-3110
The University of Edinburgh, Medical School, Fax: UK (0)131-650-6545
Teviot Place, Edinburgh EH8 9AG, Scotland. Dept: UK (0)131-650-1000

--
--
boyd@kin.ucalgary.ca