From: Michael S. Lorrey (retroman@turbont.net)
Date: Thu Mar 30 2000 - 19:49:19 MST
Mike Steven wrote:
>
> Hi all, long term lurker hereby de-lurking....
>
> Dan Fabulich <daniel.fabulich@yale.edu> wrote:
> >*How do you know which screen to look at?* There are hundreds of
> >thousands, if not millions of cameras you could be monitoring, any of
> >which may have a picture of somebody slitting a child's throat. The only
> >way to tell which one is to LOOK at all of them, requiring a gargantuan
> >amount of processing power/time. That's where the computation comes in,
> >and a hell of a lot of it.
Lets look at the number of cameras in a total ubiquitous open
surveillance system:
Range of average camera(at useful resolution): 50 m radius
Field of view: 90 degrees
Area under surveillance by 1 camera: ~7500 square meters
Area of dry surface of Earth: 148,300,000 square kilometers or,
14,830,000,000,000,000 square meters
Number of cameras necessary to cover every square meter of the dry land
on the planet:
20,000,000,000,000, or 20 trillion cameras, which if you treat each
camera as a network device on an internet like system, would require a
network a minimum of approximately 150,000 times larger than the current
worldwide internet (but only if you have nothing BUT cameras on the
network, no servers, routers, or clients), and is 20 times greater than
the capacity of the IP number system's maximum capacity. If you include
routers and clients for each of the 6 billion members of the human race,
this should add at least another 50 billion IP addresses to the network,
assuming that you can put a couple thousand cameras on one router, and
one server can handle a million or so cameras at once.
Now lets say we use Zero's Rules as a filtration protocol, recapped
here:
> > > "Notify me whenever:
> > >
> > > (1) someone accesses any of my banking data;
> > >
> > > (2) one of my listed sworn enemies (a) purchases a weapon (b) comes
> >within
> > > 200 yards of my person (c) comes within two blocks of my house (d)
> >accesses
> > > any of my data or (e) speaks my name; and
> > >
> > > (3) anyone is watching me by remote surveillance.
> >
1) seems to be pretty easy, and is more of a network/database function
than a camera function.
2) The enemies list:
a) purchases a weapon: unless you have evidence that shows a reasonable
threat of imminent harm, then I don't see how you could claim a right to
this information, however, it is easily obtainable enough by adding some
reporting functions to the NICS database system already in place. In the
instance of private sales, any video footage showing a gun in a 2 foot
proximity to an 'enemy' would be sufficient, assuming that the weapon in
question actually looks like a weapon, and that the system is capable of
delineating between dinner knives and a combat knife, a pen and a pen
gun, a hand raised in greeting and a hand poised to strike, a walking
cane or a beating cane (or a cane with concealed sword or gun), and is
able to xray a person and determine if the person has a weapon concealed
on their person...
b) comes within 200 yds of Zero's person: This is less difficult than
monitoring all cameras, so long as your system knows your GIS
coordinates at all times, and the GIS coordinates of all enemies at all
times, video feeds would only be neccessary to make sure that its
actually an enemy and not someone wearing the enemies ID chipset in an
effort to confuse and bollux up your complex defense network.
c) much like (b), only a limited number of cameras need to be
monitored.
d) accesses any of my data: This is a toughie, because you really have
no idea where in the world any of 'your data' is at any gien moment, and
you really have no control over what other people do with 'your data'. I
may query your bank info once, then resell it a million times through
encrypted channels, unless you want to do away with encryption as well
(not likely to happen).
e) speaks my name: Ok, then we are adding microphone devices to the
network, so the number of IP addy's needed is now double what I
calculated for just video feed, and you need voice recognition software
that is totally foolproof with all languages and accents and speech
impediments ( plus you'll need video sign language translators that can
read the couple dozen different sign languages), AND these applications
will need to be context sensitive to be able to discriminate between
someone actually referring to you, and someone giving a mathematics
lecture, or reading a comic book aloud...
3) Anyone who is watching you by remote surveillance: relatively
impossible, as with microtechnology, 'remote surveillance' could be any
bug, bird, etc. that doesn't need to be on your perfect network, and you
would need to be able to trace their line of sight communications
(another physical impossibility) to their own encrypted private network,
break their encryption, and track the signals to the GIS coordinates,
through any number of other devices or networks that would be stripping
and counterfeiting ID data on the packets, of one of your enemies.
>
> But given the technology required for universal surveillance, it should be
> comparatively easy for everyone to have a gizmo strapped to their wrist
> monitoring vital functions which would trigger an alarm (and nearby cameras)
> in the event of their demise. Any murderer would be caught "on tape" and
> police would converge on the location immediately. Even if some kind of slow
> poison was used, you could simply rewind the video from the victim's own
> personal camera, find the point when their food was tampered with and again,
> nail the murderer.
First you are assuming that a camera with 360 degree field of view in
both the x/y and x/z coordinate planes would have sufficient resolution
to record that info, that the persons system would not have been
pre-emptively EMPed by a person wishing to do them harm, and that the
person wishing to do them harm is either trackable by their own unit
(which they likely won't be wearing) or identifiable in video.
>
> Also, I read recently (sorry, no idea where) about software that was being
> developed (and already functional) to monitor video-surveillance in real
> time. The software triggered an alarm when it saw "suspicious" behaviour. I
> think it was still fairly primitive but I seem to remember it was already
> capable of recognising the difference between car theft and a normal driver
> getting into their car. In fact, it could even predict (presumably with a
> fair few false positives) that some one was likely to commit a crime based
> on their movements / stance etc.
There are algorithms that can discern human beings in a video feed,
however this software takes a full PC to run for ONE camera (and no
fancy gui's). You can tell by the number of times the person looks
around, as well as from their posture, whether they feel guilty about
what they are doing, but this also adds significant processing, and DOES
generate a large number of false positives, and is easy to overcome if
you know what actions the video cues on...
>
> Dan Fabulich <daniel.fabulich@yale.edu> also wrote:
> > But identifying text is a piece of cake compared to identifying people.
> > Trying to explain what a person looks like to a computer is, as Eliezer
> > might say, just one step above trying to explain it to a rock. Today, we
> > don't have anything which can do this anywhere near as well as a human can
> > (to the best of my knowledge; this project is hard, but not impossible).
> > If you'd done very much programming you'd see why this is so hard. All
> > the computer knows is a grid of numbers (a grid which, as we happen to
> > know, corresponds to a grid of colors). Based on that alone, the computer
> > has to identify what a person looks like. Notice also that the computer
> > can only use simple arithmetic and set theory to process those numbers. So
> > I want you to imagine a function which uses only arithmetic/logical
> > operators, (+, *, =, OR, NOT) and membership relations, (X is a member of
> > set Y), takes as its input a grid of numbers, and yields as its output a 1
> > if the grid of numbers corresponds to a picture of a person, and a 0
> > otherwise. Can you see how hard this is? It's not impossible, obviously,
> > but it's Very difficult. That's why we still haven't done it thus far.
>
> Actually there's a system on trial in some nightclubs in (I think) Holland
> which automatically compares photos of people who have been banned with
> incoming patrons and sounds an alarm if it recognises them. Again, this is
> based on a hazy recollection of a news article but I seem to recall that the
> system was even quite good at seeing through disguises.
The system is part of a controlled gate system, so the person always has
the same size face, at the same height and distance from the camera in
the same lighting conditions.
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:27:46 MST