Travel time is one of the most important pieces of traffic information, and various methods for measuring this parameter have been studied so far. There are some typical method, for example, the Automatic Vehicle Identification (AVI) system which measures travel time directly by reading number plates at multiple locations and a method that predicts travel time indirectly from speeds and traffic volume measured by vehicle detectors. However, both of these methods have disadvantages in installation cost and measurement precision. Therefore, the authors studied an algorithm for measuring travel time with little measurement delay and high precision by establishing a vehicle platoon correspondence using a quantified vehicle characteristic which can be measured using commonly used vehicle detectors. The evaluation results indicated an extremely good matching accuracy of 72% and RMS error in the travel time of about 2% even when vehicle height was used as the quantified characteristic. We also found that precision could be further improved by using vehicle length or combining vehicle height and vehicle length. Finally, we also propose an algorithm for improving precision by using images.
展开▼
