On 06/10/14 09:34, Bob Henson wrote:
On 05/10/2014 10:51 PM, Clive George wrote:
On 05/10/2014 21:56, Dave Liquorice wrote:
On Sun, 05 Oct 2014 20:17:02 +0100, Clive George wrote:

The missing step is automatically trawling the database looking
for clones.

You don't need to. When you do the look up you check when and
where the ANPR system last "saw" it. What you could do is
analyse the
times bewteen vehicles appearing on different cameras and build
a
"knowledge base" of how long it would normally take to get from
one camera to the ones within an area. Less than that time plus
a
margin, flag it as a possible clone.

What you're describing is automatically trawling the database
looking
for clones.

"trawling" to me is a separate process independant of the normal
lookup.

And that's almost certainly not done.

Quite likely, a waste of resources when you can do a simple
comparision during a normal look up and seta flag (and return the
result to the enquirer).

But it's the only way to pick up a cloned car if it's not otherwise
behaving suspiciously enough to trigger an individual lookup.

No you are missing the point.

No, I really am not.

Every time a vehicle passes an ANPR
camera a lookup is done to see if it's a "wanted" vehicle, when that
look up is done the date/time/location information is stored against
that registration mark. Next look up on that registration compares
where "it" is now against when/where "it" was last seen. If the time
bewteen those two locations is too short to get between them
something "odd" is going on...

Yes - and working out if the time between the locations is too short is
the trawling I'm talking about. And I don't believe that's done. It can
be done manually when a car is flagged up for some other reason - and
people have mentioned examples earlier in this thread.

What's not done is automatically working it out, and adding it to the
"wanted" list as a result.



I think you are correct, but may not be for long. The time and resources
taken to correlate all references to one number would have been so
great that, as you suggest, it would surely never have been done,
unless for, say, a specific police request.

Not at all true. I can search a database of 2.5 million postcodes in a
tenth of a second here.


on a crap PC.

On decent database and hardware its infinitely faster. Once you have a
time/gps/number plate triple, its trivial to do a search to identify
anomalies.

just don't tell anyone.



Now it would appear there
may be a very easy way to cheat the tax system, I wonder if they will
have to reconsider anyway? On the other hand it may still be too
expensive an undertaking - but if they don't do it, tax evasion may
rocket up.



--
Everything you read in newspapers is absolutely true, except for the
rare story of which you happen to have first-hand knowledge. €“ Erwin Knoll