Development of Vision-Based Vehicle Tracking for Extracting Microscopic Traffic Information

This study is to define new road capacity concept, and to develop and propose an estimation method, through the analysis of individual vehicular behaviors in continuum flow. Developments in detection technology enable various and precise traffic data collection. The U.S. HCM (Highway Capacity Manual) method does not require such various and precise traffic data, and outputs only limited results. Alternative capacity concepts, which can be classified into a stochastic model and behavioral or deterministic model, are attempts for modeling some prominent traffic flow features, namely so-called a capacity drop and a traffic hysteresis, using such various and precise traffic data. Yet, no capacity concept up-to-date can describe both features. The analysis of individual vehicular behaviors, including speed-density plot per time lap, traffic flow-speed-density diagram per each sampling interval, time headway distribution, and free flow speed distribution, is performed for overcoming the limits of the previous capacity concepts. A stochastic methods are applied to determine time headway for estimating freeway capacity for traffic control.