New system will monitor traffic in bad weather

A visual traffic monitoring system has been developed at the College of Engineering of the UAE University for vehicle detection and tracking under challenging conditions.

The system was described as highly cost-effective in a recent research report on the project, saying that it gives real-time performance.

"The system works robustly under various light conditions and cannot only count vehicles and estimate their speed, but also classify the passing vehicles using 3D measurements," said Dr Farhad Keissarian of the Department of Electrical Engineering.

The report, which was published by the Scientific Research Affairs Sector of the university, says the system has been tested on a simple computer.

"It can monitor two or three lanes simultaneously and more lanes can be processed if a faster computer is used," he pointed out.

"We present a novel approach [through the new system] to automatic traffic monitoring which focuses on two stages: vehicle segmentation and tracking and the computation of traffic parameters from the tracking data."

A two dimensional (2D) spatiotemporal (having the qualities of both space and time) image analysis and presented vehicle detection and tracking system that is destined to operate under challenging conditions.

Dr Keissarian, who is an expert in digital signal processing, said the system could handle shadows and vehicle lighting and can automatically cope with both slow and sudden light changes.

"It has been designed to recover automatically from false sensing or abrupt changes in environment on the road under its surveillance," he noted.

The report said that traffic congestion has become a significant problem in the recent years. The early solutions attempted to lay more pavements to avoid congestion but adding more lanes is becoming less feasible.

"Contemporary solutions emphasise better information and control to use the existing infrastructure more efficiently. The quest for better traffic information has increased reliance on traffic surveillance," mentioned the report.

The researchers at the Engineering College said that a vision-based approach, using video images, is promising but image processing was a problem.

Various light conditions create a need for robust algorithms which require a large amount of computational power to meet the real-time operations of a traffic monitoring system.

"Many research efforts have been made in this area, and there exist several commercial products, but there is a room for significant improvement in performance and capability of visual traffic monitoring systems," said the report.