Surveillance provides a lot of information, but it is not always used in the most intelligent way. With behaviour analysis, a sort of baseline is established that can be used to identify new occurrences.
Having a lot of cameras can give better surveillance, but it can also make the whole security picture more confusing unless the large amount of data from all those cameras can be processed.
Just looking at picture after picture sent up to the operations room by a host of cameras does not result in effective surveillance, because it is impossible for a human being to keep up with so many pictures.
So video analysis programs have been developed that can help keep watch – a sort of extra guard.
There are several ways of doing such analysis; one of them is known as behaviour analysis.
A 30-day baseline
With behaviour analysis, the system “learns” what the normal situation looks like. It learns what a typical 24-hour period looks like, not on the basis of just one day, but 30 days.
“In the course of 30 days, a baseline will be established, on the basis of which the system will react if something unusual happens”, explains Michael Struwe, project manager at Petersen-Bach, who is negotiating the behaviour analysis system AISight from BRS Labs.
“It could be a person sneaking around in the building at night. That would deviate from the normal situation, so an alarm would be given”.
The baseline, which is constructed via a neutral network, will be constantly updated.
“In a way, it’s a little bit like a bucket with a small hole. The bucket is filled up all the time with new water, but the old water at the bottom runs out as the new water comes in. It’s the same situation here. The network constantly receives new information, which it uses to update the baseline”, he explains.
Cannot be fooled
You might think that if you know about these behaviour analysis systems, fooling them should be easy. You just have to go to the same place for 30 days, and that particular behaviour, which could be preparation for burglary, will be registered as normal, so there would be no alarm when the burglary takes place because the people concerned are just part of the normal picture. But it is not that simple.
“The system can be coded so that occurrences that take place at night, for example, are never treated as part of the baseline, or that standing around certain doors or in other places where people normally wouldn’t stop does not get included in the baseline”, says Michael Struwe.
The system is particularly effective where there are a lot of cameras monitoring an area with a lot of people.
“It could be railway platforms, waiting rooms, pedestrian streets, parking lots or similar places crowded with people”, he explains.