Newsgroups: comp.robotics
Path: brunix!sgiblab!sgigate.sgi.com!olivea!spool.mu.edu!howland.reston.ans.net!paladin.american.edu!constellation!osuunx.ucc.okstate.edu!olesun!gcouger
From: gcouger@olesun.okstate.edu (Gordon Couger)
Subject: Re: Anybody know about Auto Focus
Message-ID: <1994Feb9.235102.9560@osuunx.ucc.okstate.edu>
Sender: news@osuunx.ucc.okstate.edu (USENET News System)
Nntp-Posting-Host: olesun.agen.okstate.edu
Organization: Biosystems and Agricultural Engineering at Oklahoma State Univ.
References: <2jbkajINN3sv@vision.uccs.edu>
Date: Wed, 9 Feb 1994 23:51:02 GMT
Lines: 27

In article <2jbkajINN3sv@vision.uccs.edu>,  <news@vision.uccs.edu> wrote:
>Hello all,
>
>	I'm building a robot that uses a camera to see with.  Can
>anyone give any advice about "auto focus"?  Oh, this is my first
>posting by the way so please forgive me if I make any mistakes...
>
Dr. Paul Weckler did a Phd disertation using a vidio camera and
a stepping motor to focus the lens. He did a FFT on the data and
looked for the roughest image in the frequency domain it give an
indication of contrast. Highest contrast being best focus. BTW it
was a 2d FFT. It worked very well but very computer intensive. He
then used a 2d Walch transform very simple and fast and it gave
equal results. 

All the methods I know of use either IR, or ultrasonic ranging or
make some judgment of contrast.

Hope this helps.
Gordon

/*                          Gordon Couger                             */
/*                Biosystems & Agricultural Engineering               */
/*                     Oklahoma State University                      */    
/*                114 Ag Hall, Stillwater, OK  74074                  */
/* gcouger@olesun.agen.okstate.edu 405-744-9763 day 624-2855 evenings */ 
/*         I Speak only for myself and not for anyone else            */
