Abstract
A traffic tracking software system, TrackerM, was developed by our group to detect vehicles and track their movement using video camera recordings. The study focuses on a practical analysis by using reliable data from TrackerM to propose countermeasures for avoiding a collision at a signalized intersection in the center of Niigata, Japan. Three items investigated are: short-cut behavior of subject vehicle (SV) drivers tying to turn right quickly, right-turning timing of SVs to cross the path of oncoming traffic, and countermeasures to avoid a collision between SVs and bicycles. A gap acceptance curve reveals that short-cut behaviors are liable to cause a collision particularly when a time gap is less than 4 seconds. The paper concludes with several issues for developing and improving the software system.
