Subject:
|
Re: Working motorized focus for Vision Command
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Wed, 18 Dec 2002 05:03:09 GMT
|
Viewed:
|
1083 times
|
| |
|
|
Thanks for replying!
I was thinking of the same procedure. But to help with the varried depth
problem, a small part of the viewed area could be selected to be the
focusing point.
So overall, it must be possible, right?
Timothy
In lugnet.robotics, Steve Baker <sjbaker1@airmail.net> writes:
> Bluey wrote:
> > anyone??
> >
> >
> >
> > In lugnet.robotics, Timothy Schroeder writes:
> >
> > > This may be a stupid question, but could the vision command be made to
> > > "autofocus" given special code?
>
> Well, I'm not an expert - but I imagine it should be possible to produce
> a measure of how sharp the image is. For example by adding up all the
> differences between one pixel and it's neighbours. In a sharp image,
> you'd expect a lot of nice high contrast information - which would
> result in some big jumps in brightness from one pixel to the next
> in some parts of the image. However, if the image is all blurry,
> you'd expect very gentle changes from one pixel to the next.
>
> Once you have a measure of how focussed the image is, you can then
> wind the focus dial up and down until you get the sharpest image.
>
> Of course this won't work if you point the camera at something
> blurry - or if there is a lot of depth in the scene such that you
> can't keep it ALL in focus at one time.
>
> So, I think it's possible - but I bet there are much better ways
> to get there than the technique I describe. You might want to use
> the square of the difference in pixel brightnesses for example - or
> exclude all pixels with less than a certain threshold from the testing -
> or just count the number of pixels with more than such-and-such difference
> from their neighbours.
>
> I'm sure you could find some good books on Image processing that would
> provide some tried-and-tested algorithms.
> ---------------------------- Steve Baker -------------------------
> HomeEmail: <sjbaker1@airmail.net> WorkEmail: <sjbaker@link.com>
> HomePage : http://web2.airmail.net/sjbaker1
> Projects : http://plib.sf.net http://tuxaqfh.sf.net
> http://tuxkart.sf.net http://prettypoly.sf.net
|
|
Message has 1 Reply:
 | | Re: Working motorized focus for Vision Command
|
| (...) Yes - that'll also make the algorithm run faster. (...) I'm sure it's possible. What I'm not sure about is whether the kind of naive algorithm that I could come up with after just a few minutes thought would be effective. It might take some (...) (22 years ago, 18-Dec-02, to lugnet.robotics)
|
Message is in Reply To:
| | Re: Working motorized focus for Vision Command
|
| (...) Well, I'm not an expert - but I imagine it should be possible to produce a measure of how sharp the image is. For example by adding up all the differences between one pixel and it's neighbours. In a sharp image, you'd expect a lot of nice high (...) (22 years ago, 18-Dec-02, to lugnet.robotics)
|
15 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
