Dynamic OD Estimation Simulation Optimization Based on Video License Plate Recognition

Abstract

In order to solve the problem that there are lack of research tools and methods to estimate the dynamic OD based on video license plate recognition (VLPR), by using the Visual Basic program language and component object programming technology, the dynamic OD estimation simulation platform which includes the database system, simulation model system, and data collecting and processing system was built based on micro traffic simulation model VISSIM. The shortest path algorithm, vehicle turn-ratio algorithm and BP neural network algorithm were put forward to improve the dynamic OD estimation precision. A case study indicates that (1) the efficiency and precision of vehicle turn-ratio algorithm is the best and its error is less than 5%; (2) if the OD matrix meets dynamic traffic management, the lower bound of video license plate recognition precision is 80% under the full video detector layout condition.