Hi Adam, thanks for you reply. Amazing what a few days of intensive study 
do, now I can actually understand what people in the field [of digital 
video] say. Well, 10 years ago I used to photograph a lot, so the optical 
concepts are not new to me, but digital imaging do have a few quirks that I 
were new to me.


> The two main problems you are experiencing are poor dynamic range, and
> motion blur, if I understand you correctly.

You got it right. There is also a few limiting factors such as weight (it's 
gonna be mounted on a pan/tilt base, the higher the weight, higher the cost 
of the PTB) and cost. The whole vision system (hardware) must be less than 
$2500, and I think I just got it.


> Cheap web cams will all have these problems due, primarily, to the
> lens size and light gathering capability.  The tiny lens lets so
> little light through that the sensor must "collect" light for some
> time before it's a bright enough image to use.  This causes the motion
> blur.  It also limits your dynamic range because some parts of the
> image may be saturated by the time the shutter "closes".  The sensor
> itself is usually cheap and noisy which also limits dynamic range.

I will use two firewire color CCD cameras (the boxy types), one from 
prosilica and another one from 1394 imaging. As far as lenses, I'll get two 
tamron vari-focal lenses with dc actuated iris, with minimum f-stop of 1.0, 
which is capable of great light gathering (at least theoretically). The 
option for varifocal was not a simple one, we have to spend the budget now, 
so we won't have time to find out exactly what field of view we need. With 
the varifocal lens we will have enough flexibility to postpone this 
discussion.


> When you take a regular camera which is better in both respects, you
> may be tempted to close the iris to improve depth of field - be aware
> that you're then limiting the light again and you will again pay a
> penalty in the areas you're trying to improve.  If you want everything
> to be sharp, rather than closing the iris, choose a small sensor with
> big lenses.

It's still too early to find out, but we believe that our target objects 
won't be too close to our platform, so we'll probably get away with good DOF 
even with big iris apertures (small f-stop), but if we need to control the 
iris by software, it is dc actuated.


> Two specifications that may help narrow the field down are low light
> use, and shutter speed.

That's basically what drove my specifications for lens and camera. Both 
cameras are capable of 1/100000 shutter speeds, and the lenses have minimum 
f-stop of 1.0. One of the lenses is a IR type (it focuses near IR light 
together with visible wavelenghts) that should help on low light situations, 
if it doesn't, I can always use an IR filter right? Additionally, I chose 
firewire because it simplifies cabling and no need for framegrabbers.


> If you are attempting to do stereo imaging, be aware that especially
> at high shutter speeds the cameras have to be synchronized.  This will
> be trivial, difficult, or impossible depending on the camera.

I don't believe we will, but if we want, the cameras have external 
triggering (input and output), so synchronization is [thought to be] 
trivial. One of the companies I've looked when I was researching the cameras 
is Point Grey (www.ptgrey.com), and for stereoscopic vision, they have the 
bumblebee camera that has two synch'd sensors plus a software package which 
facilitates the implementation of stereoscopic vision. They also have 
another very interesting device called ladybug, which captures 360 degree 
using 4 sensors if I'm not mistaken.

>
> Good luck!
>

Thanks!

Cheers

Padu 

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist