Date: Fri, 27 Nov 1998 16:38:28 -0800
From: kel <[EMAIL PROTECTED]>
Regarding Tom Cichoracki's 26 Nov. 1998 post on capturing images for motion
analysis in an environment containing a 165 mm diameter rotor running at 3000
rpm in a wood fiber/water suspension...
Dear Tom,
I saw your request for information and several of the responses on Spie's high
speed imaging newsgroup.
I don't fully understand all aspects of the physical layout, but I can offer
some analysis in terms of products with which I am familiar.
If we presume that we were working with a 256 x 256 pixel image and we want to
have a 30 micron object appear as 2 pixels wide (pretty much an absolute
minimum), we would need to have a fov (field of view) of no more than 256 x 15
microns = 3.84 mm.
If we want a spot *on the rotor* (see note 1 below) to appear in at least 3
images, we need to capture a frame a bit more often than every 1.3 mm (3.38
divided by 3); in other words, we need to capture 781 frames for each meter of
rotor surface that passes.
If the rotor face is moving at 26 m/s, we need a frame rate of at least 781 x 26
= just over 20,000 frames per second (fps). At least.
For the larger phenomena, the requirements are more relaxed. For a 20 mm
particle *on the rotor surface*, we might want a fov of 80 mm (limited by the
curvature of the rotor -- not by the imaging resolution). If we want the "rotor
spot" to appear in at least 3 images, we would need a new frame every 26 mm or
so (at least 38.5 frames per meter of rotor surface motion). At 26 m/s, we need
a minimum frame rate of 1000.
Although the requirements for the 30 micron objects are outside the realm of
today's low cost (< $US 30,000) high speed video systems, such products can
certainly achieve reasonable results at the 20 mm end
A few other comments:
1. If you don't expect the "phenomena" (suspended wood fibers?) to be
moving as fast as the outer surface of the rotor, your minimum frame rate
requirements may be lower.
2. Similarly, reducing your attention to a larger minimum object size could
also ease the system requirements. Doing the above analysis in reverse, a high
speed video system capable of 128 x 128 pixels at 3000 fps could track a 400
micron object in a 26 mm fov.
3. On the other hand, if we have lower image resolution (fewer pixels), we
want more detail on the object, or we want more detailed motion information
(more images of the particular object) -- especially at the 30 micron level, the
frame rate requirements go up.
4. As I wrote at the top, I am not in total understanding of your physical
layout or your view angle requirements. I have focused on a view toward the
outer surface of the rotor looking in at it from a radial orientation.
5. If the phenomena of interest are isolated and are of high contrast to
the background, "multiple exposure" type high speed imaging with at strobe
running at several times the frame rate may be beneficial. However, if this is
not true, you may find it very difficult to avoid getting a garbled mess; it's
kind of like playing "Where's Waldo" when the image is on a moving audio
turntable. Yech.
6. You may find it necessary or desirable to use a strobe (driven at the
camera's frame rate) to minimize or avoid motion blur. Fiber optic coupling
might be necessary or desirable to transmit the light through the viewport
(depending on physical configuration).
I hope these thoughts are useful. Feel free to contact me for further
information or HSV system recommendations..
Kel Tyree
http://www.time-machines.com
