University of Missouri - Columbia.
Back to Story Archive
Posted 09.08.05

MU Researcher Develops Video System to Analyze Traffic

COLUMBIA, MO -- Researchers at the University of Missouri-Columbia's College of Engineering have developed a unique video detection system involving roadside cameras that will allow agencies to more accurately analyze traffic conditions in their states. The technology will improve traffic safety and allow agencies to better plan road repairs.

Public transportation agencies increasingly are interested in the development of this type of technology, known as portable non-intrusive technology (PNIT), to replace outdated portable detectors, such as the pneumatic road tubes used for traffic surveillance. Using PNIT creates safety benefits to personnel, minimizes traffic disruptions, and is cost-effective.

"Personnel safety can be improved if personnel do not have to cross traffic lanes in order to install detectors," said Carlos Sun, MU civil engineering professor and project leader. "The high traffic demands that exist in major metropolitan areas at all hours of the day can make the deployment and retrieval of temporary in-lane detectors difficult.

"Traffic disruptions can be minimized if lane closure and the associated traffic control can be avoided. Equipment can be preserved if they are not left vulnerable in the travel way and in the path of vehicles that could target such detectors by harsh braking on top of them," Sun said.

Sun's system calls for cameras to be placed along a road, allowing the video detection system to extract various traffic parameters. A challenge that had to be overcome, though, was that the angled view the camera was providing resulted in the over-counting of vehicles since a large vehicle could encroach adjacent lanes and be counted as more than one vehicle.

In order to reduce this error and produce the most accurate vehicle volume in PNIT, Sun used a form of algebra, Boolean algebra, which combined virtual detector layouts. Sun used three detectors per lane, so that all three loops needed to be triggered at the same time in order for the vehicle to be counted. He then tested the system in 11 different locations along Interstate 70 in Missouri. Approximately 7,000 vehicles were examined during the testing.

The results showed that in general, Sun's system produced the smallest number of errors when compared with prior methods. Sun says there is still more work to be done because his system did not perform well under conditions where vehicle stoppages occurred, such as during accidents.



MU News Bureau: