An image processing system for signalized intersections

At an oversaturated intersection, most of the traffic image systems become less effective due to the combinations of driver behavior chaos and curved-shape lane geometric design. The authors describe a unique line and curve detector image process system which specifically fits in nonregular geometric layout and traffic characteristics data collections, e.g. turn ratio, queue length, volume, etc. The benefit of this image processing system is not limited to data collection, but generally can apply to intersection incident detection, signal control strategy planning and driving behavior analysis