Re: security camera software
On 26/09/14 17:44, ken wrote:
> On 09/25/2014 11:03 PM Scott Ferguson wrote:
>> On 26/09/14 01:48, Rob Owens wrote:
>>> >----- Original Message -----
>>>> >>From: "Scott Ferguson"<scott.ferguson.debian.user@gmail.com>
>>>> >>
>>>> >>I've been using motion for a few years and highly recommend it.
>>>> >>Lightweight[*1], simple, and reliable.
>>>> >>
>>>> >>Minimal configuration required (snapshot mode):- Point your camera
>>>> >>at the zone to be monitored. Take a picture. Edit the picture in
>>>> >>GIMP (mask the areas you don't want monitored for motion).
>>> >
>>> >Does this work just as a starting point for motion detection, or can
>>> >you reference this picture for the beginning of any event.
>> Motion works by comparing pictures for changes. So auto-brightness and
>> glare protection are required. Masking allows you to filter out areas of
>> the picture (representing the view of the camera) where change doesn't
>> represent the sort of movement you wish to detect. So a camera that
>> views a gate to a paddock will show a car passing through the gate -
>> it'll also show animals moving in the paddock - birds and planes,
>> clouds, passing traffic in the distance, trees and grasses. Only a small
>> rectangle in the centre of the picture is needed to detect the motion of
>> a vehicle passing through the gate - and the sensitivity of 'motion' to
>> detect changes in that area of the picture is easily tuned to filter out
>> noise (animals wandering across that part of the view etc). The end
>> result is that 'motion' will then reliably detect cars driving through
>> that gate with few or no false positives. The pictures captured are not
>> masked - the mask is only used as a reference for 'motion' to compare
>> snapshots with.
>> ...
>
> Thanks for the great info on 'motion'. Another features question:
>
> My neighborhood is overrun with squirrels and chipmunks. Is it possible
> to configure 'motion' to ignore the movement of these small creatures
> within the camera's view?
>
> How about fallen leaves blowing across the driveway?
>
>
Yes to all of the above. Did you miss the above example where I have to
deal with rabbits, birds, and moving grass in the background? :)
Another example:-
I have a client whose remote property (Nth NSW) is on the fringes of the
signal from the nearest tower. His house doesn't have line of sight to
the 3G tower and is too far for a decent signal. The solution was to
install a wireless "relay" based around the TP-Link I mentioned earlier
(about $12 from an online auction site, plus a few dollars for a cheap
USB hub, a little extra RAM, and a USB key mounted as an overlay to give
more storage space - all up about $250 including weatherproof case ,
camera, and power supply). The link is powered by a solar cell which
charges a small gel battery.
To protect the relay against theft a cheap USB webcam was added - and
'motion' is used to monitor it. The camera shows the pole mounted
TP-Link (which has line of sight on the house and the tower) and will
capture an image of anyone interfering with it. But a picture of the
thief's face would not be as useful as a picture of the number plate of
the thief's vehicle. Fortunately the camera also shows the paddock gate
through which a vehicle must enter the paddock. The paddock contains
stock (cattle and horses) as well as rabbits, birds, and grasses and
trees that move in the wind - and the occasional vehicle the travels
along the road that parallels the paddock.
So a filter was created that would only show the area where a head would
appear if someone climbed the pole and reached for the computer - and
where the front of the vehicle would be if someone stopped to open the
gate before driving into the paddock (walking would be bull-baiting and
negate the need for a camera). In this instance the built-in web server
was enabled and the port forwarded so that it could be accessed via ssh
from a public web server (secured with .htaccess) so that the landholder
can view the relay and that paddock from any location that has internet
access. (in your instance, where the system is already slow I'd suggest
disabling the webserver) - you can see the last picture in the 'motion'
image save directory.
The filter removes almost 80% of the area that doesn't need to be
monitored. Open a still shot in GIMP and use a selection tool to select
the "monitored area" - hold down shift and create additional "monitoring
areas" as needed". Add a new transparent layer then invert the
selections (now the non-monitoring areas are selected). Select the
flood/fill tool, select "fill all selected area", pick the appropriate
colour and apply. Save the filter in the appropriate format (export as...).
Now that you've excluded non-relevant areas from what is monitored just
reduced the sensitivity until only large things trigger a motion event.
Now 'motion' will ignore the squirrels, birds, blowing leaves and trees
moving in the wind. Sensitivity is a setting based on the degree of
change in the monitored area (filter part of the shot). You'll need to
make a few tests to get it right - but it's not difficult or time
consuming. Use the command-line config until you're happy then make it
permanent by adding the correct settings to the config file.
Note: 'motion' compares the last image against the reference (filter)
image - not the previous image (a failing with most infared alarms that
allows slow moving intruders to drop a bucket over the monitor).
There are a number of methods of implementing SMS alerts - the one I use
is no better than the others it's just that I have a number of Google
App accounts so it suits me to use one (actually it's Calendar app) to
send free SMS messages. gcsms[*1] makes a calendar entry, Google Apps
sends the client an SMS. Sendmail sends the pictures by email.
Relatively simple. And the client knows there's a problem, and what it
looks like, before the thief has time to leave the paddock (from where
they need to drive back past the house...).
Kind regards
Reply to: