After completing his PhD at Queen's University, Paul Miller took a job at Australia's Defence, Science and Technology Organisation. Now he's back in Belfast working to change the way CCTV technology is used to combat crime.
Every year, millions of CCTV cameras capture images of the numerous crimes committed in towns and cities across the UK. While the data they generate may eventually be used to help prosecute some offenders, it is of little value in detecting offences while they're being committed, or stopping them from happening in the first place.
That's mainly because most CCTV control rooms are flooded with information from multiple cameras, making it difficult for their operators to identify situations likely to escalate into criminal incidents.
Making CCTV a more potent weapon in crime detection and prevention is the objective of a group of researchers at the Centre for Secure Information Technologies (CSIT). Under Paul, they are developing technology capable of analysing live camera images and immediately alerting security personnel to suspicious behaviour.
Their work builds on a recently completed EPSRC-funded project known as ISIS (Intelligent Sensor Information Systems).
"Following on from that initial research, we're now working on a second multimillion-pound programme that has attracted further funding from EPSRC and a number of other partners," says Paul.
"In recent years, there has been a trend towards the convergence of physical security systems with IT security. We're taking that approach to a higher level through the use of a new computing paradigm we've developed, sensor event computing.
"This involves enhancing CCTV cameras and other types of sensors with data analysis and artificial intelligence capabilities. This makes them much more than 'dumb devices' that simply spit out raw multimedia data as current systems do. Instead, they have the ability to connect small-scale events they observe, draw conclusions about their significance and prioritise the level of threat they represent. Where this is deemed sufficiently high, an alert can be relayed to a central control room for possible human intervention.
"The level of any potential threat is assessed by each camera through a software-based scoring system that takes into account factors such as the time of day, crime statistics for the location in question and a threat assessment of the people shown.
"All of this requires a sophisticated hardware and software system architecture which we've been developing with other specialist teams at CSIT. Our work incorporates a number of firsts. These include complex gender-profiling tools for use in video analysis and the way in which we process multimedia 'events'.
"Our software will soon be made available to our membership board which includes representative from Thales and BAE Systems. They and others will then evaluate its potential for commercial development.
"We're hopeful our research could soon be incorporated in systems that could significantly reduce crime levels on our streets and public transport systems.
"Ultimately it could be adapted to protect many other kinds of critical infrastructure as well."
Click here to download a PDF version of this article.