Searching \ for 'Video Project Idea' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: techref.massmind.org/techref/displays.htm?key=video
Search entire site for: 'Video Project Idea'.

Truncated match.
PICList Thread
'Video Project Idea'
1998\10\28@210500 by Ralph Landry

flavicon
face
Hi all,
I've been trying to get the concept of a video project idea down and
would like some advice.  I'd like to use PIC to monitor a user adjustable
area of a  B/W video camera that is used in a security system.
I think what I need to do is define an area of the screen and use an A/D to
digitize the area, then if movement occurs in that area it would then either
sound an alarm or perhaps have the ability to "learn" your remote control
keys for "record" and "stop".  It could then start and stop your VCR for a
user selectable duration.

Anyone have ideas as a good PIC to use for such an application??  Or
might be interested in a joint venture project??

TIA
-Ralph
spam_OUTrlandryTakeThisOuTspamhaywood.main.nc.us
"If they call it Tourist Season why can't we shoot 'em?"
ICQ# 19545315

1998\10\28@211712 by Dave VanHorn

flavicon
face
> I think what I need to do is define an area of the screen and use an A/D to
> digitize the area, then if movement occurs in that area it would then either
> sound an alarm or perhaps have the ability to "learn" your remote control
> keys for "record" and "stop".  It could then start and stop your VCR for a
> user selectable duration.


This has been done in the past with a simple photo-resistor and
suction-cup arraingement. If the light level changes, the alarm trips.

You might monitor the entire scene, to track daylight and lighting
variations and avoid false triggering

1998\10\29@024306 by Russell McMahon

picon face
I have been thinking of something very similar but using analog
detection.
I envisaged having a row and column "window" - the detector starts
functioning at start of the column "window" in the first row in the
row window and does so on successive rows until the row at the bottom
of the row window is reached. By varying the windows' start and stop
bound you get a rectangle which can be positioned anywhere on the
screen. You then control an analog gate which sums the analog value
to produce a mean level. If anything changes substantially in the
target window the change can be detected. This is nowhere near as
potentially sophisticated as using an A2D on all data but is simpler
by far.

Line scan times are typically around 64 microseconds with some of
this unused at either end of the line - say 50uS active as rough
value for working.
To get useful A2D results you need a fastish A2D (eg 10 samples at
5uS or 50 samples at 1uS each). Worse though is that your processor
needs to handle this data rate which is a severe demand. A 20 MHz PIC
will handle 5 instructions per uS. A SCENIX at 50MHz will handle 50.

With my method a PIC would "almost" handle the gating and a SCENIX
certainly would.
(Maybe a Starfire clone goes even faster :-))
Analysis of the analog signal would be far easier (albeit less
useful) than keeping up with the data from an A2D.

You could consider using a hardware buffer and a fastish A2D and
crunching the data subsequently but you would probably be better off
with a processor aimed at data intensive processing. Some of the
modern DSPs are VERY fast and quite cheap. Maybe a 17xx PIC would
suffice.

I recently read a security magazine where they discussed video
analysis systems. Some systems will track areas of only a few pixels
to deduce direction and speed of the intruder.



regards

       Russell McMahon

From: Ralph Landry <.....rlandryKILLspamspam@spam@haywood.main.nc.us>
>I've been trying to get the concept of a video project idea down and
>would like some advice.  I'd like to use PIC to monitor a user
adjustable
>area of a  B/W video camera that is used in a security system.
>I think what I need to do is define an area of the screen and use an
A/D to
>digitize the area, then if movement occurs in that area it would
then either
>sound an alarm or perhaps have the ability to "learn" your remote
control
>keys for "record" and "stop".  It could then start and stop your VCR
for a
>user selectable duration.
>Anyone have ideas as a good PIC to use for such an application??  Or
>might be interested in a joint venture project??

1998\10\29@031410 by brad

flavicon
face
Russell McMahon wrote:
{Quote hidden}

I have used a few of these systems, from the old 4 pot adjustment (H
Top, H Bottom, V Top, V Bottom) thru to
the full colour analysing DSP based units that are supposed to work well
outdoors.
If you want to bounce some ideas off me, I'll be happy to help where I
can.
I have an old analog (I say old, cause it's very old technology) unit
downstairs now. I'll rip the cover off
and have a look inside.


--
-----------------------------------
Brad Campbell
Technical Manager

Seme Electrical Engineering Co
59 Collingwood St Osborne Park 6017
Western Australia
Ph    :-+61 8 9445 2577
Fax   :-+61 8 9244 1327
Email :- bradspamKILLspamseme.com.au

 /"\
 \ /
  X  ASCII RIBBON CAMPAIGN
 / \ AGAINST HTML MAIL

1998\10\29@092454 by Harrison Cooper

flavicon
face
I'm somewhat involved in video, so I know a little bit about this stuff
(designing a video frame buffer right now...no PIC's tho)

First off, to convert the video.  Probably need to use a Flash A/D,
depending on how good you want, but probably 8 bits would be fine.  6
bits may be a little less desirable.  How often do you want to capture
and compare data?  How much error do you want to live with?  Video
coming in is interlaced.  Do you want or need to capture both fields, or
is one sufficient? I would imagine that only one field would work fine.
OK, so now, how much error can be stood before you want to trigger.

So, you need two frame buffers, to hold the current frame and the prior
frame.  Then, just do a fast compare of data.  If you get a difference,
then turn on the recorder.

Sounds pretty simple, but its not.  The analog section and digitizer
needs to be in sync with the video coming in.  You need to be able to
write to the frame buffer.  During the vertical sync period, or after
you have captured the portion of video you are intersted in, need to
read from both frame buffers and do the fast compare, write to an error
register, then decide if that error, or amount of errors warrant the
turning on of the recorder.  Then be ready to swap buffers and take the
next frame in.  You might be able to map the errors such that if you get
a significant amount in a certain address block that indicates movement.
Or store the errors over several fields and then see if the error
remains (nothing can move that fast), then perhaps it was a valid error.

Bottom line.  I don't think a PIC would work, except for maybe the input
controller. But if it was my project, I'd do the thing in some fast CPLD
or FPGA.  5ns clock to Q parts are now available in either flavor.

Just my opinion, but then again, that is what this forum is for right?

1998\10\29@132100 by Peter L. Peres

picon face
On Thu, 29 Oct 1998, Harrison Cooper wrote:

> I'm somewhat involved in video, so I know a little bit about this stuff
> (designing a video frame buffer right now...no PIC's tho)
>
> So, you need two frame buffers, to hold the current frame and the prior
> frame.  Then, just do a fast compare of data.  If you get a difference,
> then turn on the recorder.

No-one stores the whole frame, let alone two of them to boot, unless there
is another reason for this. One digitizes video at 10 MHz and faster and
hashes the area of interest in a PLD or ASIC for every other frame or
field. The comparison is on the hash. How exactly the hashing, the
sampling and decimation are done determines what kind of detail is dropped
and how big the error to trigger etc.  Hire a good mathematician, there
are interesting stunts to be done like this ;)

Peter

1998\10\29@132107 by Peter L. Peres

picon face
On Thu, 29 Oct 1998, Russell McMahon wrote:

> I recently read a security magazine where they discussed video
> analysis systems. Some systems will track areas of only a few pixels
> to deduce direction and speed of the intruder.

This has applications that are totally different from security systems and
it should not be widely available.

Peter

1998\10\29@154440 by Sean Breheny

face picon face
At 07:46 PM 10/29/98 +0000, you wrote:
>This has applications that are totally different from security systems and
>it should not be widely available.

Are you implying missile technology again?

>
>Peter
>
+-------------------------------+
| Sean Breheny                  |
| Amateur Radio Callsign: KA3YXM|
| Electrical Engineering Student|
+-------------------------------+
Save lives, please look at http://www.all.org
Personal page: http://www.people.cornell.edu/pages/shb7
.....shb7KILLspamspam.....cornell.edu  Phone(USA): (607) 253-0315 ICQ #: 3329174

1998\10\29@163358 by Andy Kunz

flavicon
face
At 03:43 PM 10/29/98 -0500, you wrote:
>At 07:46 PM 10/29/98 +0000, you wrote:
>>This has applications that are totally different from security systems and
>>it should not be widely available.
>
>Are you implying missile technology again?

Probably Phreaking.  I have a neat story about a demo Vice President George
Bush received on the base I worked at.  The bad guys could do it without
entering restricted space.  very bad.

Andy


==================================================================
Andy Kunz - Statistical Research, Inc. - Westfield, New Jersey USA
==================================================================

1998\10\30@062005 by Peter L. Peres

picon face
On Thu, 29 Oct 1998, Sean Breheny wrote:

> At 07:46 PM 10/29/98 +0000, you wrote:
> >This has applications that are totally different from security systems and
> >it should not be widely available.
>
> Are you implying missile technology again?

Why 'again' ? When did I do that before ? I've nothing to do with missiles
or the army, else I'd not be talking here. What do you think ?

What I am implying is, that the method and its algorythms belong in the
Waasenaar treaty list for now, and had better stay in there. If for no
other reason, then because I did not like the Scuds in 1991. Get it ?

Peter

1998\10\31@030200 by Russell McMahon

picon face
Slightly further OT but ...

Our phone system differs substantially from that in US & Canada and
was, allegedly immune to blue boxing, unlike the former two. BUT, a
tech from Canada came to live here and worked for the Post Office, as
did I. He demonstrated that with certain older exchanges you could
get a circuit into the heart of the system by dialing a toll line and
releasing it and then reseizing it just right - took some skill.
Then, if you knew the format, and he did, you could not only dial
anywhere for free but also charge the call to any number in the
country. Clearly only certain charges would be undetectable. I saw
this done so it was not just a boast.
I mentioned this in passing to one of our organisations upper echelon
men and he was so adamant that it was impossible that I didn't pursue
it. The tech concerned only did this as a party trick, (I think)
All the old exchanges (step by step with group selectors et al) are
gone now so its safe to tell this tale.

From: Andy Kunz <EraseMEmtdesignspam_OUTspamTakeThisOuTFAST.NET>
>Probably Phreaking.  I have a neat story about a demo Vice President
George
>Bush received on the base I worked at.  The bad guys could do it
without
>entering restricted space.  very bad.

More... (looser matching)
- Last day of these posts
- In 1998 , 1999 only
- Today
- New search...