I'm thinking such information no matter how compressed, it need not even be
graphical in nature.
I'm wondering if the statistical analysis of such things in some sort of alphanumeric/integer form is possible, image recognition tossing the pixels away, and generating a string that is shorthand for: "
Car (small, red) went westbound on City Avenue, north on HY55, off at 12th st. Parked at Lat/Long XYZ for 12hrs 15 minutes"... the powers that be could data-mine the HELL out of that sort of thing.
Then ignore all the cars moving about in a "regular" way, people regularly commuting, local residents running errands etc. could be filtered out completely with software. And "anomalous cars/people" moving about in ways not already found to be commonplace in the database stick out like a sore thumb algorithmically. A little cross check with RFID tollway systems, and some traffic cameras/red-light cameras, and accuracy, at least in high density urban areas could approach 100%
I'm thinking when it's "time for the revolution", your cell having access to a delivery truck or something that's always making regular rounds in your AO, and moving people/materials along said route, with those stops in between,
will be a really good idea.