Newsgroups: comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!usenet.eel.ufl.edu!news-feed-1.peachnet.edu!gatech!swrinde!pipex!lyra.csx.cam.ac.uk!warwick!nott-cs!lut.ac.uk!usenet
From: "A.Torelli-94" <elat3@lut.ac.uk>
Subject: Clarity about stereo matching
Sender: usenet@lut.ac.uk (Usenet-News)
Message-ID: <D4nKzp.BFn@lut.ac.uk>
Date: Mon, 27 Feb 1995 09:57:25 GMT
Nntp-Posting-Host: aragorn.lut.ac.uk
Organization: Loughborough University, UK.
Lines: 20

Hi all, I'm Andrea.

ABOUT STEREO MATCHING:
	We assume that the camera model is Known, so that epipolar 
lines can be computed.Assumes horizontal epipolar lines, corresponding
to parallel image planes.

NOW:
	The searchingfor candidate matches is restricted to one
dimension.The algorithm attempts to match only nonhorizontal lines, 
becouse the disparity of horizontal lines is SUBJECT TO LARGE ERROR.

I WOULD LIKE TO KNOW:
	Why the horizontal lines,that are parallel to the epipolar
lines, are subject to large error?????

Thanks in advance

name:ANDREA TORELLI
e-mail:elat3@lut.ac.uk
