Subject:
|
Re: Vision command + linux
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Sun, 14 Dec 2003 20:27:40 GMT
|
Original-From:
|
PeterBalch <peterbalch@compuserve.comSAYNOTOSPAM>
|
Viewed:
|
981 times
|
| |
|
|
Steve
> What I did was to divide the image into small regions (4x4 pixels each)
> and find the 16 highest contrast regions in the image. Then, I take
> two consecutive frames and try to find a match for each of those regions
That must be fairly costly if you want bigger offsets and probably fails
during a zoom or roll of the camera.
> It also (currently) has a tendency to latch onto vertical
> or horizontal edges and to slide along them randomly.
Makes sense.
If you convert the image into (straight or curved) line segments then its
easier and cheaper to match "corresponding" segments between images.
My guess is that "Snakes" would help here. I used Snakes for an "AI" vision
project last year (analysing medical ultrasonic images) and was very
impressed with how well they coped with noisy blurred images. Their
"capture distance" is far better than the 4 pixels you're working with.
They're easy to implement too.
If you can somehow get snakes fitted to various features on one image (e.g.
by dropping random snakes on the first image and attracting them to
features) then apply the same snakes to the next image and see how they
move, you should get a good estimate of what the camera did.
If the image is mainly of straight-edged geometic objects then there's a
very old technique called "Shape Attractor" that's very quick at finding
how to map one scene onto another.
Peter
|
|
Message has 2 Replies:
 | | Re: Vision command + linux
|
| (...) Do you have a web site or something where I could read more about this? ---...--- Steve Baker ---...--- HomeEmail: <sjbaker1@airmail.net> WorkEmail: <sjbaker@link.com> HomePage : (URL) : (4 URLs) GEEK CODE BLOCK----- GCS d-- s:+ a+ C++++$ (...) (21 years ago, 14-Dec-03, to lugnet.robotics)
|
3 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
