Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!aplcen!uunet!seismo!beno!black
From: black@beno.CSS.GOV (Mike Black)
Newsgroups: comp.graphics
Subject: Re: Algorithm to detect moving obect.
Summary: Intensity variations are detectable
Keywords: image processing
Message-ID: <49082@seismo.CSS.GOV>
Date: 6 Sep 90 22:43:06 GMT
References: <8010@helios.TAMU.EDU>
Sender: usenet@seismo.CSS.GOV
Organization: Center for Seismic Studies, Arlington, VA
Lines: 34

In article <8010@helios.TAMU.EDU> hyeom@cs.tamu.edu (Heon Y Yeom) writes:
>Forgive me if my question is something trivial.
>We have a video tape recorded on a highway from the side of the road.
>I need to somehow identify the vehicles passing by without somebody watch the
>tape for couple hours.
>I have a relatively simple frame grabber which is able to capture video data
>60 times per second with 64 grayscale.  What I'd like to do is to move the
>video to a series of image files and process them.
>Is there an algorithm to detect moving obect from a series of images ?

What you want to do Heon is basically a form of frame-to-frame comparison.
What you do could depend on what your image hardware is capabile of but would
involve this:

1.  Since you are working with a static scene a car entering the field of
view will cause a variation in intensity in the image.  A dark car would
decrease the intensity and a light car would increase it.  You can subtract
all the pixels in one frame from the last frame and examine them for a
certain amount of change. i.e. 100 pixels varying more than 10 might work.

2.  This involves a lot of number crunching for an image.  You may be able
to undersample the image to reduce the processing.  For example, if a car
occupies 1/3 the field-of-view on a 512x512 pixel image, you could examine
every 10th line and every 10th pixel and reduce your processing by two
orders of magnitude.  It simply depends on how large the car is in the image.

3.  Your image hardware may already have some of these functions.  If you
wish to converse further, please e-mail and I'll help you out.
Mike...
--
-------------------------------------------------------------------------------
: usenet: black@beno.CSS.GOV   :  land line: 407-494-5853  : I want a computer:
: real home: Melbourne, FL     :  home line: 407-242-8619  : that does it all!:
-------------------------------------------------------------------------------