🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: sci.image.processing
Article #389

Re: Discriminating Objects from their shadows.

#389
From: David A. Honig
Date: Fri, 28 Aug 1992 15:16
18 lines
1347 bytes
 whb@castle.ed.ac.uk (H Bruce) writes:
>Can anyone point me to work done in the area of discriminating between objects
>and the shadows they cast ?
>
>I need this as shadows are causing havoc with my object tracking algorithm.
>

Look into some of the computational perception approaches to edge-finding.
(Marr, Canny, Hildreth...) A shadow will likely have a rather broad edge,
vs. an occlusion (object) edge, due to the penumbra.  Also if you have color
info the color changes in shadows will be constrained in ways (eg, RGB all
reduced by an equal fraction, if certain assumptions about light sources are
true) that may allow you to discriminate them from other edges.
--
David "Dances with Motion Sensors" Honig

'People that are really very wierd can get into sensative positions
and have a tremendous impact on history.'	-- Vice President Dan Quayle
Technical Details
Message-ID: <2A9E4359.24318@ics.uci.edu>
Path: arc1-ca.pugleaf.net!sparky!uunet!news.claremont.edu!ucivax!honig
References: <24945@castle.ed.ac.uk>

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!news.claremont.edu!ucivax!honig
From: honig@ruffles.ics.uci.edu (David A. Honig)
Subject: Re: Discriminating Objects from their shadows.
Nntp-Posting-Host: ruffles.ics.uci.edu
Message-ID: <2A9E4359.24318@ics.uci.edu>
Newsgroups: sci.image.processing
Reply-To: honig@ics.uci.edu (David A. Honig)
Organization: UC Disneyland, in the Kingdom of Bren
Lines: 18
Date: 28 Aug 92 15:16:41 GMT
Distribution: sci
References: <24945@castle.ed.ac.uk>