From owner-freebsd-hackers Mon Feb 5 15:58:22 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id PAA07731 for hackers-outgoing; Mon, 5 Feb 1996 15:58:22 -0800 (PST) Received: from seagull.rtd.com (root@seagull.rtd.com [198.102.68.2]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id PAA07725 for ; Mon, 5 Feb 1996 15:58:20 -0800 (PST) Received: (from dgy@localhost) by seagull.rtd.com (8.6.12/1.2) id QAA09358; Mon, 5 Feb 1996 16:58:11 -0700 From: Don Yuniskis Message-Id: <199602052358.QAA09358@seagull.rtd.com> Subject: Re: motion detection? To: langfod@maui.com (David Langford) Date: Mon, 5 Feb 1996 16:58:11 -0700 (MST) Cc: freebsd-hackers@freefall.FreeBSD.org (FreeBSD hackers) In-Reply-To: <199602052300.NAA12351@ maui.com> from "David Langford" at Feb 5, 96 01:00:09 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@FreeBSD.ORG Precedence: bulk > >Does anyone have a program to detect motion and then take a snapshot > >of the scene? > > A cheap way (computation wise) would be to have a microwave or infrared > detector trigger a signal for the sytem to take snapshot(s). > > If you want to do it in software the simplest way would be to hack > something like what MPEG uses to find differences between frames. > > True motion detection of course would require you to find object boundaries. > > I wonder if you used edge detection and check if the edges themselves > changed... > > So lets see, Your watching TV on your monitor, someone walks down the driveway > an X-window pops up to alert you to the fact and turns on the X-10 light > switch by the front door. You notice its your wife (etc..) and you click the > little icon labeled stereo and the classical CD starts playing and the lights > dim. Or am I reading too much into all this :) If you are looking for really *gross* triggers, just look at average light level changes over the entire frame. And, since the number of samples (i.e. pixels) is constant, this would just be a sum of all dots. This needs a tiny filter on the average to ignore small changes as would be evident from quantization and then a discriminator to trip your alarm. hey, I *said* it would be gross but I imagine you could implement this in an hour and test it just as quick.