Re: CCTV/PC's

[Nigel Orr <Nigel.Orr@xxxxxxx>]

At 22:58 02/05/00 +0100, you wrote:
>It's possible to down sample video images and compare frames to detect
>movement directly. My MEng was in image processing so I have books on
>this in the loft, but no time to start messing about with 3x3 mask
>functions!

It works quite well on inside pictures, but it's harder to do
outside.  Inside, I've had a reliable system working by simply subtracting
one frame from the next, and setting a threshold for the sum of the
absolute pixel differences (you can then also display the 'difference'
image, to see what has caused the triggering).

It's trickier outside, with different areas continuously brightening and
darkening at different rates due to shadows and reflections.  You really
need to do proper movement detection then, which is where mathematicians
wake up and get to work and us engineers doze off at the sight of large
matrices and funny symbols <g> .  The PIRs with built in cameras
aren't
really the answer, as the PIR is most sensitive to movement across the
sensor, so you tend to get pictures of the sides of people's faces...
Ideally you would want internal cameras pointing at the room door from one
corner, and a PIR in another corner where it is most sensitive to someone
coming in that door.

>I vaguely remember a chip camera for the BBC Model B that could do
>this- the Beasty perhaps? Anyone still got copies of Acorn User with a
>type-in listing of such software ;-)

Vlsi Vision (http://vvl.co.uk) were
proposing a chip camera with onboard
processing to do the same thing, but I'm not sure what's happened to that
recently.

Nigel


------------------------------------------------------------------------
eGroups eLerts
It's Easy. It's Fun. Best of All, it's Free!
http://click.egroups.com/1/3863/5/_/2065/_/957341138/
------------------------------------------------------------------------