"Depth of focus" is the range at which an object is focused by a lens.
Most Webcams focus from a couple inches (maybe 6-8") to infinity. I'm
not sure this is what you're looking for here, but maybe you could
For general information about computer vision and track planning see
this link, for starters (and follow the other links):
Well I think I know exactly what you need. If you're not familiar with
Nuts and Volts Magazine, and Servo Magazine, I highly recommed you
check them out, they have tons of excellent articles on all aspects of
robotics and electronics from beginner through to advanced concepts.
There is one particluar set of articles by Robin Hewitt which was
published over 5 issues of Servo in 2005 called 'The Gift of Sight'. In
the articles Hewitt goes through all of the math and theory behind
using a single camera for calculating depth and locating objects in
space, and works step by step through an example application. He also
goes into detail on how to create a 'home' marker that the robot can
recognize and navigate to. These articles probably explain everything
you need to at least give you a good grounding and some starting
applications. The articles are in the July through November 2005
issues. You can purchase back issues for $5 each here:
http://www.servomagazine.com/store.php?page 05, or you can also
purchase a CD of back issues for $30 however it doesn't look like they
have 2005 on CD available yet.
Even if you do buy the print issues it would only be $25 and the other
articles are definitely worth the price. I guess I sound like a shill
for them but I've been subscribed to both magazines for 2 years now and
I have learned a slew of new concepts and technologies that I otherwise
would not have been exposed to in online robotics forums.
These articles should be what you need, however one note: in order to
calculate depth locations in the manner they do in the article you must
have images from two different locations, so you may need to construct
a mechanism for moving the camera locations. I think you can calculate
using images that are only 6 inches apart so perhaps just a moving
slider with the camera mounted to it. Although you may be able to
calculate depth if you have some sort of visual cues in your frame of
vision to reference relative to the target object. Good luck!
A simple method is to mount a laser pointer directly above or below the
camera. Scan the center column of the image to determine where the
laser hits the target. Then use trig to calculate the distance. Even
easier, construct a look-up table by placing a target at known distances
and measuring the pixel location. The look-up table would probably be
more accurate (harder to mess up, and camera images usually have
Plus it looks cool.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.