Faster than a Speeding Car

R. Winn Hardin

Americans love their cars. But the feeling of freedom associated with a full tank of gas has turned into a dread of mobile prisons for many urbanites as traffic jams turn daily commutes into daily crawls. In June and again in October 1996, Washington state transportation officials turned to a machine vision system to help plan a solution to traffic congestion on the Tacoma Narrows Bridge.
Rebuilt in the 1940s, the bridge was designed to handle 40,000 vehicles per day. Today, nearly twice that number cross the span, causing traffic backups and delays during peak travel times. Should Washington expand the bridge, build a new gateway or increase the numbers of ferries to Tacoma and Seattle? To answer these questions, transportation officials needed data on driving patterns such as origins and directions of travel, average speeds and final destinations.
Because placing observers on the bridge to manually record license plates, dates, directions and times was not feasible on the four lanes crossing the Tacoma Narrows, officials turned to an automatic license plate reader from Computer Recognition Systems Inc. of Cambridge, Mass., and Transformation Systems of Houston. By suspending black-and-white cameras above each lane and wiring them to a mobile trailer containing four NRS2 license plate reading systems and manual viewing stations, operators recorded during two sessions 500,000 license plates on vehicles traveling at speeds in excess of 50 mph. The NRS2 systems use calculation-intensive image processing algorithms to identify each plate within the frame in real-time. Then the special-purpose modules decide whether an image is coherent enough for further processing and review.
Transportation officials mailed a survey to the owner of each recognizable plate -- with a goal of 50 percent of the total group -- to find out the travel patterns of Seattle/Tacoma commuters. Initial results from the surveys are not yet available.
While using machine vision to monitor traffic has been around since the 1970s, advances in the technology of imaging have expanded applications to include traffic flow analysis, stolen vehicle detection, speed and toll enforcement, access control for major arteries and vehicle emission enforcement. Using license plate readers as a means to track traffic flow has proven successful in the UK and in the US, including Florida, Texas, West Virginia, Nevada and New York.