Determining Origin-Destination Flows Across a Two-Intersection Network from Non-Overlapping UAV and Ground-Based Imagery

Origin-Destination (OD) flows are important inputs to both off-line planning and real-time traffic management applications. Typically OD flows are estimated from time-consuming and expensive travel surveys or license plate matching studies. These estimates can be updated using link-based observations. The use of Unmanned Aerial Vehicles (UAVs) to monitor traffic networks introduces the possibility of directly determining OD flows form small networks. A methodology that combines non-overlapping imagery is developed and applied to a two-intersection network. The empirical findings show that even with short imaging time, interesting and nontrivial results can be obtained.