You are in:
- EDITORIAL CONTENT » News » News
How to harness the power of cameras and sensors
How to fully harness in times of crisis the combined power of the thousands of cameras and sensors that populate the land? That is the question that DYVINE (Dynamic Visual Networks), a project funded by the EC’s Sixth Framework Programme, has been endeavouring to answer since September 2006. The answer is due in a few months time, and the solution could be highly attractive for contingency responders all around the world.
DYVINE acknowledges that visual sensors such as cameras are both affordable and widely used. However, at times of crisis – when access to these sensors would actually be useful – they are often not accessible by the people who need them the most because they are stand alone systems often owned by private companies or separate agencies.
The 12-partner consortium that forms DYVINE aims to support the risk management cycle via the integration of thousands of fixed/mobile/airborne video sensors, providing a resilient communications system that will support their use in real time.
Project Co-ordinator Philippe Chrobocinski is confident that the outcome of DYVINE will shortly be on offer to the world, either as a whole solution or in segments, depending on the legacy systems already in a city or area. “We will be pitching this to many different parties: government – mostly for border control and security systems, City halls for Urban Security; but we want to talk to people responsible for critical infrastructure, airports, ports, nuclear plants.”
Philippe however makes it clear that DYVINE is a flexible solution that can be used both during major disasters and day-to-day emergencies and citizen security.
To Philippe’s knowledge, DYVINE and the sheer number of amalgamation capabilities being researched are unprecedented, which is why 12 partners were necessary to form the consortium.
Among the partners; EADS is designing the IT architecture; Martec (France) is developing the interface between the sensors and the system; Realviz (France) specialises in image processing; Surrey University is working on wimax integration; Valencia University on body worn cameras integration; and EPFL (Switzerland) on 3D processing for locating objectives, amongst others.
There have been three main challenges. First of all the building of an architecture capable of integrating into one system a large number of cameras – there could as many as four thousand in a city. Secondly, the ability to integrate new cameras such as body worn cameras or new sensors into the system during times of crisis. “The third challenge has been to optimise the video processing to allow real time processing.”
A successful trial has already taken place in Segrate, Italy, in November last year, demonstrating common operational picture, video stitching, and video compression in an urban scenario with an important truck accident generating huge troubles in the city.
In the first step, a traffic jam was detected automatically via live video analysis and data fusion. The truck accident event was detected and shown to the operator; a subsequent fire on the truck was then also detected and reported; patrol car attendance shown on the system; and live images transmitted by the first responder were also shown on the system. The fusion and correlation of all the alarms were demonstrated on the overall picture.
In the third step, a hazardous area was outlined, demonstrating how the system could define a security perimeter, detect entry points; and observe moving people and objects in the perimeter through a tracking system.
“The trial proved that DYVINE can be deployed easily and can interface with legacy surveillance systems without major impact on its architecture. The demonstration in Italy reached all the objectives with a realistic scenario – mostly with real events and data – in which the system operated seamlessly and with the expected results.”
The invited operational end-users of the trials successfully validated the general concept of DYVINE, added Philippe, and provided valuable inputs for the next phases, in terms of adaptability to the various agencies way of working and in terms of system operability.
A second trial was due to take place this year in Miraflores, Peru – a location chosen because it is situated in an area prone to earthquakes – but unfortunately it had to be cancelled for political reasons. However, the final DYVINE trial is due to be carried out in Valencia, Spain, in July this year, the culmination of the two-year project.
The trial in Valencia will demonstrate all the technical capabilities of DYVINE in a scenario covering all the possible configurations of the system. Starting with day-to-day operation by the city traffic management, it will escalate after a disaster to a full configuration integrating all the sensor networks and additional mobile cameras to provide an accurate real-time picture to the responders (Police and Civil Protection).
After that, it will be time to take DYVINE to the open market, and Philippe believes that having been designed to be highly affordable, it will be attractive. “Our aim was to develop an affordable model to install on top of existing systems. Obviously if there are no existing camera networks it would be expensive to install. If we are talking about installing smart layers inside a fully equipped city then it is not costly.”
Various configurations are available depending on the national contingency structure. Valencia city hall, for example, is in charge of all civil protection agencies, so in this case the city hall would be the owner of the system. “In the case of France it is more complex because the local police have very limited responsibility, so here perhaps it would be the national police.”
Spain would be a good example of a suitable end user, because common emergenc
y centres have been developed here for 112 emergency calls. “In the typical Spanish 112 room you have the legacy systems of the police, ambulance and fire. You can think of DYVINE as the layer that combines all that information and synthesises it.” In a real scenario, it could mean for example a fireman being able to access live video footage from the cameras of a commercial shopping centre. “The idea for this type of customer is that they can deploy selected parts of DYVINE to enable connection to CCTV.”
All has not been smooth, adds Philippe, and some challenges still remain. Data protection is still a problematic issue. “We had to take into account both European and national laws. The use of systems that can help identifying a person are subject to strict legal conditions. And secondly, connecting different agencies together means that you should be able to control what is done with each one of the entering video footage, originally managed by these distinct agencies.
“You would need within the system some control to ensure that all data is exploited in the right way. This area was not in the DYVINE framework for development but an exhaustive study has been performed by KU Leuven to identify all the relevant issues that need to be addressed when implementing such a system. Some solutions are suggested, but the main headache is that the data protection legislations have been enacted 10 years ago and are thus not adapted to systems such as the one developed by DYVINE. Specific CCTV legislations adopted. in some countries does not deal with the specific problems raised by such systems neither.”
