Automated and robust traffic surveillance system for Europe

We may not fully appreciate it when commuting to and from work via motorways, bridges and tunnels, but road traffic surveillance systems play a key role in mitigating congestion and enabling emergency services to respond to accidents in an effective and rapid manner. E! 4160 VICATS, a successful collaborative research effort sponsored by EUREKA, has yielded highly promising results in the development of an innovative surveillance system which requires minimal human input. The new software could transform existing systems, improve robustness and establish a new paradigm in road traffic monitoring.

The project, coordinated by Dr Vladimir Crnojević at the University of Novi Sad, Serbia, has been founded upon an official cooperation between said university and University College Ghent, Belgium, and has also drawn on the skills and expertise of two industrial partners working in the field, Fitis-JU (Serbia) and Traficon N.V. (Belgium). Such interface between research and industry has proven crucial in the application of the software, which uses computer algorithms to develop real-time assessment models for traffic management.

Limitations of existing methods

Until now, video surveillance systems have been heavily reliant on intensive human input and concentration; footage captured on closed-circuit television (CCTV) systems is directed to centralised surveillance control centres, where it is studied and analysed by individuals. Although technological advancements have ensured that CCTV installation is economical, the financial costs of employing individuals to analyse footage continuously are substantial. Yet, more pointedly, the extent of information that must be processed by humans demands a great deal of concentration, which, regrettably, means that errors and misjudgements can be -- and indeed are -- made on occasion, with potentially fatal consequences.

While recorded CCTV output is often of high value in retrospectively examining the circumstances leading up to a critical scenario, the partners involved in E! 4160 VICATS have developed a robust and automated software package capable of producing intelligent and reliable traffic assessments based on the trajectories of moving vehicles. The system anticipates potential hazards or congestion, before relaying a signal to humans, directing them to particular scenarios and enabling them to react accordingly in a prompt fashion. In this light, the project has engendered enormous hope for preventing congestion, identifying accidents that do arise, and ensuring rapid emergency responses are deployed when necessary -- benefiting, in essence, road users across the EU.

Autonomous and intelligent

Remarkably, the traffic monitoring system developed under the E! 4160 VICATS initiative is completely autonomous, gathering information on traffic conditions 24 hours a day, without the need to be initialised by a human operator. Speaking of the software's robustness and ability to gather and analyse intelligent data, Crnojević comments: "The system can operate successfully in all weather conditions and is capable of deducing a wealth of high-level knowledge. It can detect pedestrians, traffic queues and accidents; classify vehicles and measure the volume of traffic on a stretch of road; calculate the average speed of moving vehicles; and recognise speed limit violations." Owing to the open architecture of existing surveillance systems, it is anticipated that the software developed and optimised under the project's banner will be assimilated seamlessly into current traffic monitoring technologies, bearing particular relevance to tunnel management.

While the past decade has seen a preponderance of research efforts focused on the development of sophisticated computer algorithms, the majority of models that have emerged perform well only in controlled laboratory conditions. Drawing on this, Crnojević comments: "The sudden change of light in real-world scenarios can ruin the whole process of image analysis that has been designed in laboratories, and this can occur quite commonly in real situations." In overcoming this significant challenge, results have indicated that the software developed through the course of the project can cope with extremely difficult lighting conditions that are manifest in tunnels. This marks a major breakthrough, and the system has also recorded positive results in surveying traffic on bridges and at crossroads.

Fostering collaboration

EUREKA has been pivotal in enabling the E! 4160 VICATS project to coordinate its research; it has fostered collaboration between researchers and industrial partners within the consortium, blending theoretical knowledge with long-standing experience in the field of video surveillance. "Without the EUREKA label and funding," Crnojević states, "this project would not have been possible. EUREKA has connected two European research institutes with international industry partners, and has facilitated the exchange of ideas which has culminated in two publications: one has been published in the journal, Lecture Notes in Computer Science, while the other will be submitted very soon."

Notably, the E! 4160 VICATS project has successfully managed to transfer research from the laboratory and apply it to real-world surveillance. As part of the project's dissemination programme, demos and results which demonstrate the robustness of the new software are freely accessible to view on YouTube (http://www.youtube.com/watch?v=VIfM8RI4xAI&feature=player_embedded) and the project's website (http://www.ursusgroup.com).

Crnojević reflects on the advancements the project has delivered: "Our software can be considered an upgrade to the surveillance system previously offered by our industrial partner. It is a continuation of improving the reliability and robustness of a system that can operate in very difficult conditions."