RE: USB linux webcam recommendations

Replies:

Parents:

Ted,

It was really early in the morning when I sent you my reply. I didn't intend
to keep it
off the fogo list so I'll put it on now. I hope that's ok with you?

Ok, here's my lame idea about capturing movement.

640x480 bitmap picture can be cut up into n squared pieces and to keep it
simple let's cut the picture into 4 X 4 = 16 smaller squares thus 1/16 would
represent
a 40x30 bitmap.
           640
+--------+--------+--------+--------+
|        |        |        |        |
|        |        |        |        |
|    1   |    2   |    3   |    4   |
|        |        |        |        |
+--------+--------+--------+--------+
|        |        |        |        |
|        |        |        |        |
|    5   |    6   |    7   |    8   |
|        |        |        |        |
+--------+--------+--------+--------+ 480
|        |        |        |        |
|        |        |        |        |
|    9   |   10   |   11   |   12   |
|        |        |        |        |
+--------+--------+--------+--------+
|        |        |        |        |
|        |        |        |        |
|   13   |   14   |   15   |   16   |
|        |        |        |        |
+--------+--------+--------+--------+


Now instead of diff'ing only one image of 640x480
you would diff 16 images of 40x30.

you'll have to tweak the diff percentage a bit say 10%
for leaf wind rustling and just image noise which is acceptable changes.

[let's leave time of day for later. (a) historical samples
can smooth this out better so we need to keep some time of day
samples around. (b) or we sample often that the differences in light
will be negligible. (c) we can create  a light table where we predict
colour changes based on lighting dimming or brightening. ]

If there is significant changes(more than 10%) in section 13, 9, 5
to 14, 10, 6 we can say the object is moving from
left to right and someone should be alerted about this movement.

same for top to bottom 1, 2, 3, 4 changed to 9, 10, 11, 12

Maybe it's too simple but that's the principal I always use KISS, and it
works great with the wife too.! :-)

Cheers,

David.

-----Original Message-----
From: Ted Guild [mailto:[email protected]]
Sent: Monday, August 26, 2002 9:15 AM
To: David A. Jones
Subject: Re: USB linux webcam recommendations



We could follow up on fogo but since you kept this off the list, I
will too.

"David A. Jones" <[email protected]> writes:

> Ted,
>
> Hi I saw your request for a cheap camera for around the house, so you
could
> watch the
> UPS guy get shredded by your huge vicious German Shepard a.k.a man eater
:-)
> Ya, I'm
> sure you think he's just being cute playing with the mail man like a rag
> doll. Next you'll
> tell me he's loving an affectionate but don't bug him today because he in
a
> bad mood after eating
> the next door neighbor ;-). Ok, that's enough I know.

Honestly your honor he was just showing the mail man his
affection as you can see in this mpeg I have...

> Well I always see this add popping up annoying the hell out of me (when
I'm
> trying to download some porn ;-) he he)
> (just kidding) about this wireless web cam so I thought I'd pass this
along
> to you.
>
> http://www.x10.com/products/x10_vk45a.htm
>
> is a good place to start for a $79 web cam that you can pan around.

Thanks.  I'll check them out.  I've seen some other panning/zoom
cameras but the ones I saw were pretty expensive.  I figured capturing
the movement with something like motion [1] was a fair first step, I
could software zoom somewhat on 640x480 images and still have some
clarity.

> There are other packages that you can assemble too if you want something
> more.

Panning could be added separately, the monitor guy has some custom
circuitry and code [7] for following the motion, panning based on the
direction it takes - or so I surmised it is suppose to do.

So you honor as you can see in this next segment my dog playfully
chased him around the car in the driveway and then dragged him back
into full view in front of the car.

> A wireless web cam maybe better suited to your needs. Look mom, no
> wires! ;-)

I was initially thinking that route but the cameras I came across
were often ~400USD.  I'm doing the 802.11b wireless thing around the
house so it would fit in well.  I do have an old laptop that I might
be able to get a pcmcia usb card for and save a few hundred if I use
it and a cheap webcam.

I'm seeing some wireless ones that run off of battery as well.  What
would be cool would be to hook in a solar panel and then never have to
change the batteries - probably would need to change them at least
every other week otherwise.  Then I'd just find the tallest tree in
the area and get it up there for a nice 360 bird's eye view and hope
it stands the weather.

> Hope this helps

It does.  Your suggestions are very much appreciated.

> Cheers,
>
> David.
>
> P.S.
>
> If you need some software on sensing movement I have an idea or two about
> doing
> that in a program that I could easily write.

I'm interested in your idea here.  I recall some time back some image
file format (but which?) that worked well with standard UNIX diff.  If
that's the case then you'd only need to count up the number of lines
of diff output and at a certain threshold consider there to be
movement beyond the rustle of leaves in the wind, clouds casting
shadows and such.

> 1. http://motion.technolust.cx/
7. http://motion.technolust.cx/tracking

--
Ted Guild <[email protected]>
http://www.guilds.net

Re: USB linux webcam recommendations

Replies:

  • None.

Parents:

"David A. Jones" <[email protected]> writes:

> Ted,
>
> It was really early in the morning when I sent you my reply. I didn't intend
> to keep it
> off the fogo list so I'll put it on now. I hope that's ok with you?

Absolutely.

> Ok, here's my lame idea about capturing movement.

Not lame at all, similar to one or more approaches I noticed.  Another
is where you overlay an image of nulled out space, eg where the trees
are so that you only care about movement in the rest of a certain
restricted area.  With your grid approach you could simply ignore
certain grids, say 1, 5 and 4.

> 640x480 bitmap picture can be cut up into n squared pieces and to keep it
> simple let's cut the picture into 4 X 4 = 16 smaller squares thus 1/16 would
> represent
> a 40x30 bitmap.
>             640
> +--------+--------+--------+--------+
> |        |        |        |        |
> |        |        |        |        |
> |    1   |    2   |    3   |    4   |
> |        |        |        |        |
> +--------+--------+--------+--------+
> |        |        |        |        |
> |        |        |        |        |
> |    5   |    6   |    7   |    8   |
> |        |        |        |        |
> +--------+--------+--------+--------+ 480
> |        |        |        |        |
> |        |        |        |        |
> |    9   |   10   |   11   |   12   |
> |        |        |        |        |
> +--------+--------+--------+--------+
> |        |        |        |        |
> |        |        |        |        |
> |   13   |   14   |   15   |   16   |
> |        |        |        |        |
> +--------+--------+--------+--------+
>
>
> Now instead of diff'ing only one image of 640x480
> you would diff 16 images of 40x30.
>
> you'll have to tweak the diff percentage a bit say 10%
> for leaf wind rustling and just image noise which is acceptable changes.
>
> [let's leave time of day for later. (a) historical samples
> can smooth this out better so we need to keep some time of day
> samples around. (b) or we sample often that the differences in light
> will be negligible. (c) we can create  a light table where we predict
> colour changes based on lighting dimming or brightening. ]

To handle time of day darkening you could compare the diffs of all the
grids simultaneously and see if they are universally changing as that
would mean:

a. it's getting dark out
b. some really large like an intergalactic spaceship is covering the
whole field of vision - in other words they're landing right on top of
us and your webcam, computer and everything else in the area will be
squashed anyway.

Clouds casting moving shadows will be tough.

> If there is significant changes(more than 10%) in section 13, 9, 5
> to 14, 10, 6 we can say the object is moving from
> left to right and someone should be alerted about this movement.
>
> same for top to bottom 1, 2, 3, 4 changed to 9, 10, 11, 12
>
> Maybe it's too simple but that's the principal I always use KISS, and it
> works great with the wife too.! :-)

Sometimes the most elegant software is the most simple.

--
Ted Guild <[email protected]>
http://www.guilds.net

HURL: fogo mailing list archives, maintained by Gerald Oskoboiny