I'm working on a security application that detects a moving object (usually
a human ...) in an outdoor space measuring approx 200 x 500 ft. Output are
x/y (maybe even z) coordinates, which are fed to a x/y/z- mount with a
video camera on it. Mount gets coordinates, aims, zooms in, focuses -
smile!. Video camera and x/y/z mount are done. All I need is the ranging
module. Ultrasound is out of the picture because of the distances involved.
Motion detectors etc. don't work, because I need the exact location of the
object, not just presence detection.
Bertel Schmitt wrote:
>
> I'm working on a security application that detects a moving object (usually
> a human ...) in an outdoor space measuring approx 200 x 500 ft. Output are
> x/y (maybe even z) coordinates, which are fed to a x/y/z- mount with a
> video camera on it. Mount gets coordinates, aims, zooms in, focuses -
> smile!. Video camera and x/y/z mount are done. All I need is the ranging
> module. Ultrasound is out of the picture because of the distances involved.
> Motion detectors etc. don't work, because I need the exact location of the
> object, not just presence detection.
>
> Any ideas?
sure , use an ir laser at 36khz
detect it with a remote receiver
when you scan the whole area like a
sweep the system has to know
where it is pointing - if reflection
comes - bingo !
one source is enough to get the angle .
two to get the exact location
At 11:58 AM 1/20/98 +0100, Bertel Schmitt wrote:
>I'm working on a security application that detects a moving object (usually
>a human ...) in an outdoor space measuring approx 200 x 500 ft. Output are
>x/y (maybe even z) coordinates, which are fed to a x/y/z- mount with a
>video camera on it. Mount gets coordinates, aims, zooms in, focuses -
>smile!. Video camera and x/y/z mount are done. All I need is the ranging
>module. Ultrasound is out of the picture because of the distances involved.
>Motion detectors etc. don't work, because I need the exact location of the
>object, not just presence detection.
>
>Any ideas?
>
What are the details of the space?
Is it enclosed by walls on all four sides?
Can equipment/sensors/etc. be mounted anywhere along the perimeter?
How about within the space?
Is the space empty, or are there obstacles that would cause "blind spots"?
What resolution is needed -- how accurately does one need to locate the
target?
Is it possible to mount some sort of device to the target? (Not likely, as
you mentioned security, but I thought I'd ask...)
In any case, here's a possible solution, assuming the following:
- Space is normally empty
- Sensors may be placed along perimeter, but not in center
- Position of target needs to be accurate within 3 feet or so
- Position updates needed every few seconds.
Place triggerable xenon flash units behind IR filters every 10 feet along
both long walls. Place IR sensors every foot or so along both long walls.
Trigger each flash in sequence, recording which sensors do *not* see which
flashes. Lines drawn between flashes and the sensors "in the dark" will
intersect the target. Simple algebra calculates intersection point for
each pair of readings.
- Rick "Impractical Ideas Ya Us" Dickinson
+---------------------------------+---------------------------+
| Enterprise ArchiTechs Company |"You can't reason someone |
| Lotus Certified Notes | out of a position they |
| Appl. Design & Administration | didn't reason themselves |
|(818)563-1061 spam_OUTrtdTakeThisOuTnotesguy.com | into" -- Rick Adams, |
| http://www.eArchiTechs.com | in alt.folklore.urban |
+---------------------------------+---------------------------+
> I'm working on a security application that detects a moving object (usually
> a human ...) in an outdoor space measuring approx 200 x 500 ft. Output are
> x/y (maybe even z) coordinates, which are fed to a x/y/z- mount with a
> video camera on it. Mount gets coordinates, aims, zooms in, focuses -
> smile!. Video camera and x/y/z mount are done. All I need is the ranging
> module. Ultrasound is out of the picture because of the distances involved.
> Motion detectors etc. don't work, because I need the exact location of the
> object, not just presence detection.
>
> Any ideas?
>Place triggerable xenon flash units behind IR filters every 10 feet along
>both long walls. Place IR sensors every foot or so along both long walls.
>Trigger each flash in sequence, recording which sensors do *not* see which
>flashes. Lines drawn between flashes and the sensors "in the dark" will
>intersect the target. Simple algebra calculates intersection point for
>each pair of readings.
Interesting, and I thought about it (I'm prone to brute force solutions).
But since we need x/y info, these sensors (or beams) would have to be
placed at the short _and_ the long walls, which would make it a bit
cost-prohibitive. Also, the ground is sloping.
To answer you questionnaire:
>What are the details of the space?
It's a yard. Fences on both sides, one side waterfront (and hence wide open
to intruders).
>Is it enclosed by walls on all four sides?
Fences on long sides, house on short side, wazterfront at the end.
>Can equipment/sensors/etc. be mounted anywhere along the perimeter?
Sure, within limits.
>How about within the space?
No.
>Is the space empty, or are there obstacles that would cause "blind spots"?
More or less empty. But trees are growing. There's a pool. People tend to
set up pool furniture in summer. Shrubbery grows. You get the picture.
>What resolution is needed -- how accurately does one need to locate the
>target?
For a good close-up shot ... I figure 1ft
>Is it possible to mount some sort of device to the target? (Not likely, as
>you mentioned security, but I thought I'd ask...)
At 02:26 PM 1/20/98 +0100, charley wrote:
>
>sure , use an ir laser at 36khz
>detect it with a remote receiver
>when you scan the whole area like a
>sweep the system has to know
>where it is pointing - if reflection
>comes - bingo !
>one source is enough to get the angle .
>two to get the exact location
Interesting. Bewen thinking along similar lines, but couldn't find
references. Could you point me towards some implementation data?
>> I'm working on a security application that detects a moving object (usually
>> a human ...) in an outdoor space measuring approx 200 x 500 ft. Output are
>> x/y (maybe even z) coordinates, which are fed to a x/y/z- mount with a
>> video camera on it. Mount gets coordinates, aims, zooms in, focuses -
>> smile!. Video camera and x/y/z mount are done. All I need is the ranging
>> module. Ultrasound is out of the picture because of the distances involved.
>> Motion detectors etc. don't work, because I need the exact location of the
>> object, not just presence detection.
>>
What about some image processing to detect the movement using the camera
you already have? (I guess it depends on the Field of View)..
PC based frame grabbers are 'relatively' cheap now days, and the motion
detection algorithm could be simple frame by frame comparison, execution
times should not be too much of a constraint. Once you have determined the
center of movement you could then zoom/autofocus and track..
> At 09:08 AM 1/20/98 -0800, Rick Dickinson
>
> >Place triggerable xenon flash units behind IR filters every 10 feet along
> >both long walls. Place IR sensors every foot or so along both long walls.
> >Trigger each flash in sequence, recording which sensors do *not* see which
> >flashes. Lines drawn between flashes and the sensors "in the dark" will
> >intersect the target. Simple algebra calculates intersection point for
> >each pair of readings.
>
> Interesting, and I thought about it (I'm prone to brute force solutions).
> But since we need x/y info, these sensors (or beams) would have to be
> placed at the short _and_ the long walls, which would make it a bit
> cost-prohibitive. Also, the ground is sloping.
The flash method, while maybe not good for lifetime/cost affectiveness,
only needs flashes along the long wall.
F\ Sensor
\
F \ Sensor
Intruder
F / Sensor
/
F / Sensor
/
F / Sensor
/
F Sensor
Of course, speed of flashes (which could change when an intruder showed
up would determine how accurate your spotting is.) would determine
how fine you could pinpoint your intruder. As should be obvious to a
casual observer flashes have to fire in sequence.
Actually, with this method I don't think you'd really need to worry about
an intruder. Who would go into this space with all the flashes lighting
up? :)
If you could deal with the control issue (maybe a rotating flat mirror)
you could do the same thing with two lasers near the midpoint (for
greatest accuracy) or near the base (or building) for most safety.
Problems with these systems, reflected light, rain. As for static
things in the way that is easily programmed around. At 8PM (or whenever
it is supposed to be empty) take your snapshot of which sensors are
contacted.
>
>At 09:08 AM 1/20/98 -0800, Rick Dickinson
>
>>Place triggerable xenon flash units behind IR filters every 10 feet along
>>both long walls. Place IR sensors every foot or so along both long walls.
>>Trigger each flash in sequence, recording which sensors do *not* see which
>>flashes. Lines drawn between flashes and the sensors "in the dark" will
>>intersect the target. Simple algebra calculates intersection point for
>>each pair of readings.
>
>Interesting, and I thought about it (I'm prone to brute force solutions).
But since we need x/y info, these sensors (or beams) would have to be
placed at the short _and_ the long walls, which would make it a bit
cost-prohibitive. Also, the ground is sloping.
>
>To answer you questionnaire:
>
>
>>What are the details of the space?
>
>It's a yard. Fences on both sides, one side waterfront (and hence wide
open to intruders). {Quote hidden}
>
>
>>Is it enclosed by walls on all four sides?
>
>Fences on long sides, house on short side, wazterfront at the end.
>
>
>>Can equipment/sensors/etc. be mounted anywhere along the perimeter?
>Sure, within limits.
>
>>How about within the space?
>
>No.
>
>>Is the space empty, or are there obstacles that would cause "blind spots"?
>
>More or less empty. But trees are growing. There's a pool. People tend to
set up pool furniture in summer. Shrubbery grows. You get the picture. {Quote hidden}
>
>
>>What resolution is needed -- how accurately does one need to locate the
>>target?
>
>For a good close-up shot ... I figure 1ft
>
>
>
>>Is it possible to mount some sort of device to the target? (Not likely, as
>>you mentioned security, but I thought I'd ask...)
>
>No :)
>
>BS
>
>
>At 02:26 PM 1/20/98 +0100, charley wrote:
>>
>>sure , use an ir laser at 36khz
>>detect it with a remote receiver
>>when you scan the whole area like a
>>sweep the system has to know
>>where it is pointing - if reflection
>>comes - bingo !
>>one source is enough to get the angle .
>>two to get the exact location
>
>Interesting. Bewen thinking along similar lines, but couldn't find
references. Could you point me towards some implementation data?
>
>BS
At 07:31 AM 1/21/98 +1000, Ray Gardiner wrote:
>
>What about some image processing to detect the movement using the camera
>you already have? (I guess it depends on the Field of View)..
>
>PC based frame grabbers are 'relatively' cheap now days, and the motion
>detection algorithm could be simple frame by frame comparison, execution
>times should not be too much of a constraint. Once you have determined the
>center of movement you could then zoom/autofocus and track..
Place is too large and method too slow. Camera doesn't have total area in
field of view, so it needs to pan-compare-etc. Not practical.
The ideal inexpensive human detector is a pyroelectric sensor,
Eltec has some nice units and some arrays that might be useful. The
thing to remember about pyroelectric sensors is that they only detect
changes in temperature. I think a couple of the pyroelectric arrays
combined with a lens would also work pretty well with a minimum of
processing overhead. Not sure about the range issue though.
I'm working on a security application that detects a moving
object (usually
a human ...) in an outdoor space measuring approx 200 x 500 ft.
Output are
x/y (maybe even z) coordinates, which are fed to a x/y/z- mount
with a
video camera on it. Mount gets coordinates, aims, zooms in,
focuses -
smile!. Video camera and x/y/z mount are done. All I need is the
ranging
module. Ultrasound is out of the picture because of the
distances involved.
Motion detectors etc. don't work, because I need the exact
location of the
object, not just presence detection.
I've got an Idea. Choose a sensor that is capable of deciding position with some
uncertainty (doesn't matter if it is kind'a lousy), actually you should be able
to use
a couple of cameras with oo-focus. Weighting together the positions
from different cameras assuming gaussian errors, you should be able to get a
very good
measure of the exact position of the object.
Other ideas?
--
-------------------------------------------
BassLab
Production, Mix, Remix facility
Mail us : KILLspambasslabKILLspamusa.net
Call us : +46-(0)708-833331
-------------------------------------------
That's exactly what we are looking for (and money doesn't seem to be too
much of an object ...) but the range of that thing is only 10 meters =
30ft. We need more than 10 times that.
> The ideal inexpensive human detector is a pyroelectric sensor,
>Eltec has some nice units and some arrays that might be useful. The
>thing to remember about pyroelectric sensors is that they only detect
>changes in temperature. I think a couple of the pyroelectric arrays
>combined with a lens would also work pretty well with a minimum of
>processing overhead. Not sure about the range issue though.
They currently have nice long range motion sensors in the yard. Narrow
beam, I'd say 300 ft range. Trouble is they are a bit fussy, sometimes get
triggered by a large bird or low sun.
>In any case, here's a possible solution, assuming the following:
> - Space is normally empty
> - Sensors may be placed along perimeter, but not in center
> - Position of target needs to be accurate within 3 feet or so
> - Position updates needed every few seconds.
>
>Place triggerable xenon flash units behind IR filters every 10 feet along
>both long walls. Place IR sensors every foot or so along both long walls.
>Trigger each flash in sequence, recording which sensors do *not* see which
>flashes. Lines drawn between flashes and the sensors "in the dark" will
>intersect the target. Simple algebra calculates intersection point for
>each pair of readings.
>
> - Rick "Impractical Ideas Ya Us" Dickinson
A' la lawnmowers and goats ....
Hire good defence lawyer.
Substitute .22 firearms for each xenon flash unit. Forget about the IR
sensors. Count the arrival of lead opposite the array of firearms. If less
than fired, collect the intruder ...notify kin.
>>It's a yard. Fences on both sides, one side waterfront (and hence
>wide
>open to intruders).
>>Fences on long sides, house on short side, wazterfront at the end.
Think like a burglar (though it's often a mistake to think that burglars
will act rationally). They aren't going to be too interested in hanging
around the perimeter of the lawn and shrubbery. They'll try to break
into the house, steal cars or boats, and maybe jump in the pool. So
concentrate the motion detectors and detail camera(s) there. Eventually
they will wander into these areas and get close enough to be identified.
Along the lines of your original plan, consider using a fixed-mounted
camera with a wide angle lens near the moveable one. Process the image
from the fixed camera to detect any changes and direct the telephoto
camera there. When nothing seems to be happening, scan the telephoto
camera around in a fixed pattern for a close look at valuable objects or
areas.
At 09:13 PM 1/20/98 +0100, Bertel Schmitt wrote:
>At 09:08 AM 1/20/98 -0800, Rick Dickinson
>
>>Place triggerable xenon flash units behind IR filters every 10 feet along
>>both long walls. Place IR sensors every foot or so along both long walls.
>>Trigger each flash in sequence, recording which sensors do *not* see which
>>flashes. Lines drawn between flashes and the sensors "in the dark" will
>>intersect the target. Simple algebra calculates intersection point for
>>each pair of readings.
>
>Interesting, and I thought about it (I'm prone to brute force solutions).
>But since we need x/y info, these sensors (or beams) would have to be
>placed at the short _and_ the long walls, which would make it a bit
>cost-prohibitive. Also, the ground is sloping.
Actually, no need to have them along the short walls. The flashes will
radiate in *all* directions, and will theoretically illuminate all sensors
simultaneously (subject to lightspeed constraints :-) except where blocked.
Intersections of multiple flash-shadow lines should pinpoint almost any
point in the area....
>To answer you questionnaire:
>
>>What are the details of the space?
>
>It's a yard. Fences on both sides, one side waterfront (and hence wide open
>to intruders).
>Fences on long sides, house on short side, wazterfront at the end.
Where is the camera? I assume it's probably by the house (seems like the
best location to me).
>>Can equipment/sensors/etc. be mounted anywhere along the perimeter?
>Sure, within limits.
>>How about within the space?
>No.
>>Is the space empty, or are there obstacles that would cause "blind spots"?
>
>More or less empty. But trees are growing. There's a pool. People tend to
>set up pool furniture in summer. Shrubbery grows. You get the picture.
>>What resolution is needed -- how accurately does one need to locate the
>>target?
>
>For a good close-up shot ... I figure 1ft
Camera focus is pretty forgiving at large distances, and somewhat finicky
up close. Sounds to me like accuracy would be most important near the
camera, and less important in the distance.
>>Is it possible to mount some sort of device to the target? (Not likely, as
>>you mentioned security, but I thought I'd ask...)
>
>No :)
Okay - time for impractical idea #2.
Place the camera in the middle of the short side by the house. Place 2
additional cameras (these can be el-cheapo fixed-focus black & white
cameras, as we are not looking for quality, just using them for ranging) in
the corners at the ends of the short sides, aiming at the opposite corner
of the rectangle. Digitize their outputs to provide two independent
bitmaps, with probably 8 bit greyscale for each bit.
For each bitmap, perform the following manipulations:
- "Smooth" each pixel by summing up the surrounding 8 pixels,
shifting right 3 bits (dividing by 8), adding the original
value, and shifting right again (dividing by 2).
- Compare this smoothed image with the last smoothed image by
subtracting pixel values and taking the absolute value.
- "Sharpen" the difference image by taking each pixel value,
shifting left 3 times (multiply by 8), adding the original
value, and subtracting each of the surrounding pixel values.
Find the brightest pixel, and that is the location of greatest movement in
that frame.
Calibrate the sensors by placing a flashing light at a particular location
in the yard. Each position will result in a different set of sensor
coordinates for each screen. Store the values measured in a lookup table,
then use interpolation between measured locations to approximate the
location of any movement in the yard.
The calibration idea can be extended if you add a means to manually adjust
the camera positioning and focus. Each calibration point can be entered by
simply setting up your flashing light, aiming and focusing the camera at
the flashing light, and hitting a button to lock in the setting. With
enough calibration points, camera pointing and focus settings can be
directly interpolated from the sensor data lookup tables.
- Rick "What, Me Impractical?" Dickinson
+---------------------------------+---------------------------+
| Enterprise ArchiTechs Company |"You can't reason someone |
| Lotus Certified Notes | out of a position they |
| Appl. Design & Administration | didn't reason themselves |
|(818)563-1061 spamBeGonertdspamBeGonenotesguy.com | into" -- Rick Adams, |
| http://www.eArchiTechs.com | in alt.folklore.urban |
+---------------------------------+---------------------------+
At 10:21 PM 1/20/98 +0100, Bertel Schmitt wrote:
>
>Interesting points. Thanks for the clarification. I don't think the owner
>will want to turn their property into a landing strip (BTW, a major airport
>is 5 miles away, which could lead to interesting phenomena at night) so
>strobes with I/R filters would be called for :)
>
>BS
Actually, in my impractical design idea #1, I suggested:
>Place triggerable xenon flash units behind IR filters every 10 feet along
^^^^^^^^^^^^^^^^^
So, I actually was thinking along those very lines....
- Rick "Next week's lesson: Reading for Comprehension" Dickinson
+---------------------------------+---------------------------+
| Enterprise ArchiTechs Company |"You can't reason someone |
| Lotus Certified Notes | out of a position they |
| Appl. Design & Administration | didn't reason themselves |
|(818)563-1061 TakeThisOuTrtdEraseMEspam_OUTnotesguy.com | into" -- Rick Adams, |
| http://www.eArchiTechs.com | in alt.folklore.urban |
+---------------------------------+---------------------------+
Bertel Schmitt wrote:
>
> >sure , use an ir laser at 36khz
> >detect it with a remote receiver
> >when you scan the whole area like a
> >sweep the system has to know
> >where it is pointing - if reflection
> >comes - bingo !
> >one source is enough to get the angle .
> >two to get the exact location
>
> Interesting. Bewen thinking along similar lines, but couldn't find
> references. Could you point me towards some implementation data?
no , this is an idea which was
originally for a no-limits
resolution microscope .
I heard about it at the medical
university it worked great
and broke the maximum resolution minimal
wavelength optical density
rule , which limits the orig.microscopes
.
but I'd suggest the siemens sfh infra
receiver
it demodulates and amplifies well .
Bertel Schmitt wrote:
>
> >sure , use an ir laser at 36khz
> >detect it with a remote receiver
> >when you scan the whole area like a
> >sweep the system has to know
> >where it is pointing - if reflection
> >comes - bingo !
> >one source is enough to get the angle .
> >two to get the exact location
>
> Interesting. Bewen thinking along similar lines, but couldn't find
> references. Could you point me towards some implementation data?
no , this is an idea which was
originally for a no-limits
resolution microscope .
I heard about it at the medical
university it worked great
and broke the maximum resolution minimal
wavelength optical density
rule , which limits the orig.microscopes
.
but I'd suggest the siemens sfh infra
receiver
it demodulates and amplifies well .
Why not detect the intruder as they cross the perimeter, you could use
multiple light barriers to detect the entry point and focus the camera
on that. Alternately use a light barrier to detect intrusion and then
scan/record the yard for a period of time before reverting to detection
mode. Or for the really brave of hart, sink sensors in the ground and
use induction comparison i.e. like the surface of a touch screen. The
latter is probably not practical due to the effects of weather.
If cost is not the issue, I'd recommend multiple cameras capable of
detecting changes in the scene.
BTW, ever thought of an electric fence or a dog :-)
Gareth
Project & Operations Manager, STB Systems Inc.
Tel: +44 1232 258402, Fax +44 1232 319970, WWW Page: http://www.stb.com
> Interesting, and I thought about it (I'm prone to brute force solutions).
> But since we need x/y info, these sensors (or beams) would have to be
> placed at the short _and_ the long walls, which would make it a bit
> cost-prohibitive. Also, the ground is sloping.
Here's a shot :
How about taking two of those laser pointing devices and pointing each
at a polished spinning axle with a flat surface on one side. You need
a few sensors along the walls. You know when a beam should hit the
sensors. A look-up table could do the rest.
--
Friendly Regards
Tjaart van der Walt RemoveMEtjaartTakeThisOuTwasp.co.za
_____________________________________________________________
| WASP International http://www.wasp.co.za/~tjaart/index.html |
| R&D Engineer : GSM peripheral services development |
| Vehicle tracking | Telemetry systems | GSM data transfer |
| Voice : +27-(0)11-622-8686 | Fax : +27-(0)11-622-8973 |
| WGS-84 : 26¡10.52'S 28¡06.19'E |
|_____________________________________________________________|
My summary:
1) Your space is anisomorphic; if I pick the word correctly; that is, it
is not of uniform texture. Almost any scanning modality will of
necessity detect irregularities.
2) Probably the only useful modality to distinguish important objects
(humans in particular) by characteristic is heat, and that only useful
at night.
3) The alternative is image comparison. The "reference" image may be
either fixed or updated by some delay (low-pass filter) mechanism.
4) A mixture of the above would seem most appropriate. PIR detectors,
as already installed, are the simplest implementation of this, but give
a view with no spatial dimension (ie., only time!).
5) What is therefore called for is a two or possibly three-dimensional
implementation. The simplest way to do this to my mind is an a small
array of linear (one- dimensional) resolvers strategically placed. Such
devices are readily constructed using PIR sensors and drum scanners,
both readily available in the "disposals" (surplus) market and of
course, new.
6) Processing of the data stream from these is probably best performed
centrally. Each sensor requires a "reference" store, some processing to
separate differential data and (low-pass) update the reference, and a
mapping function of its linear "view" into (part of) the actual area in
conjunction with overlapping sensors. It would be appropriate as
suggested, to implement a "teaching" function of the fully implemented
system to map critical points and subsequently interpolate between
these. The map would become an array of point entries, each documenting
an actual grid location, an identity/ position (one-dimensional) pair
for each sensor which can "see" this location, and an az/ el pair to
direct the camera to this location.
The algorithm to interpolate (map) readings is most challenging. It
goes along the lines of: If one sensor only registers, approximate by
averaging (least squares) those reference points with the closest
readings, giving preference to points ONLY seen by that sensor. If two
or more sensors register, use points with entries for those sensors
accordingly.
Not trivial. It is either formal or informal "fuzzy" logic!
>How much would a RADAR cost to cover the area? How easy is it to
>detect a person on RADAR?
That reminds me..
A few years back there were some press releases regarding
a "radar on a chip" system developed by lawrence livermore labs.
I seem to recall that they were going to be manufactured for
automotive anti-collision systems etc.. might be worth checking!
I believe a lot of licences have been sold, maybe some of these
licencees should have reached commercial product status by now!..
>How much would a RADAR cost to cover the area? How easy is it to
>detect a person on RADAR?
That reminds me..
A few years back there were some press releases regarding
a "radar on a chip" system developed by lawrence livermore labs.
I seem to recall that they were going to be manufactured for
automotive anti-collision systems etc.. might be worth checking!
I believe a lot of licences have been sold, maybe some of these
licencees should have reached commercial product status by now!..
Aided by the input of various list members, I decided to go the two camera
route to detect motion, well, actually humans moving in an area of 200x500
feet. While doing research on image processing libraries, I found the
following interesting link:
Bertel, the recent issue of EDN (15 Jan 98) mentions two low cost
software packages for motion-sensing. One package, "Gotcha" by
Prescient Systems allows you to set parameters that trigger a camera and
computer's action. For example, how much motion is necessary to start
recording or take an action. You can also mask areas of a scene to allow
motion. The package costs $44.95 and is suppose to work with any camera
that produces video for Windows. Their web site is at:
>Aided by the input of various list members, I decided to go the two camera
>route to detect motion, well, actually humans moving in an area of 200x500
>feet. While doing research on image processing libraries, I found the
>following interesting link:
>
>http://www.faceit.com/devkit/api.html
>
>
>The library actually finds _faces_
>
>BS
> What about some image processing to detect the movement using the camera
> you already have? (I guess it depends on the Field of View)..
>
> PC based frame grabbers are 'relatively' cheap now days, and the motion
> detection algorithm could be simple frame by frame comparison, execution
> times should not be too much of a constraint. Once you have determined the
> center of movement you could then zoom/autofocus and track..
>
> Maybe, some cross-hairs on the display.......!!
>
And what about stereoscopics?? I mean, maybe using two cameras, each one
rotating to center the subject, and mesuring the angular rotation of each
camera you could calculate the position of your moving subject.
notesguy.com | into" -- Rick Adams, |
