PDA

View Full Version : Re: Camera vibrations



Ton Van Den Bogert
05-04-1999, 05:00 AM
I am forwarding this message for Jim Walton .
For some reason he was unable to post directly to the list.

-- Ton van den Bogert, Biomch-L co-moderator

>From: jim@4DVideo.com
>Subject: Re: Camera vibrations
>To: Patrick Castagno
>Cc: BIOMCH-L@NIC.SURFNET.NL
>X-Mailer: SPRY Mail Version: 04.00.06.21


Patrick:

In regard to mounting cameras on vibrating structures ...

this is a fairly common issue, I've dealt with it many times when
evaluating vibration in mechanical systems.

You can "register" the raw images by including 2 (or more) stationary
(retroreflective?) reference targets on a fixed structure(s) beyond
your calibrated region. (A wall/the floor?) Just make sure you
distribute these reference targets as far apart as possible.
As usual, redundancy improves the performance of the procedure.

Before you perform any data manipulations (DLT, etc.), shift and rotate
the location of every image to the common (2-D fixed) reference frame
defined by the marks on the wall. It's not difficult to write a simple
filter to do this using simple trigonometry. But don't forget, you'll
also need to perform an identical shift and rotate on your calibration
data.

The process described above will work even under the most gruelling
conditions--with cameras attached to vibrating machinery, for example.

However, I suspect your colleague's problems are far less severe.
In fact, he might want to measure the motions of some "reference"
targets to determine the extent of his problem. If the reference
targets are fixed to a far wall and the cameras are fixed, the images
of the reference targets should also remain stationary. If the vibration
of the cameras is significant, the images of the reference targets
will appear to move. If the time-histories of these motions are
plotted, you can then measure their amplitude, direction and frequency
content. (Bear in mind the frequency data may be aliased, but for the
most part, the fans/motors in HVAC systems introduce fairly
low-frequency noise.)

If you know enough about the geometry of the camera/wall configuration,
you can estimate how the vibration might impact any data recovered at
points between the camera and wall. Finally, if he (your colleague)
writes a filter to correct the problem, he can use the same "test" data
to check that his program actually DOES eliminate the noise introduced
by the vibrating attachment. (It's always a good idea to check first!)

I hope this helps. I'm posting this to the list, in general, because
this is a fairly common problem, and others may want to use the
technique.

Regards,

Jim WALTON
Chairman, SPIE Working Group on High-Speed Photography,
Videography and Photonics

************************************************** ***********
************************************************** ***********
* *
* 4D VIDEO CELEBRATED ITS 10TH ANNIVERSARY ON OCTOBER 1 '98 *
* *
************************************************** ***********
************************************************** ***********
* * *
* JAMES S. WALTON, Ph.D. * INTERNET: Jim@4DVideo.com *
* President, 4D VIDEO * *
* 825 Gravenstein Hwy North * PHONE: 707/829-8883 *
* Suite #4 * *
* SEBASTOPOL, CA 95472 * FAX: 707/829-3527 *
* * *
************************************************** ***********
************************************************** ***********

---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------