Methodology of OD Matrix Estimation Based on Video Recordings and Traffic Counts

Abstract

Microscopic traffic modelling as the input data for traffic could use two types of data 1) OD matrices; 2) input intensities for each entering point in the network. The goal of the paper is to describe in detail and demonstrate application of the methodology of OD matrix evaluation based on video recording and traffic counts done in the study area. The approbation of the methodology was done in Riga city in the area which covers part of few city districts. The study area was surrounded by the video recording devices. In the same time manual counting of the traffic was performed. Next the manual decoding of the video recording was performed, recording the type of the vehicle and licence plate number for each recording site with split on incoming and outgoing traffic. This gave the information about origin and destination transport zones of the traffic flow and allows to construct initial OD matrix. The evaluated initial OD matrix was used to estimate the probability matrix, which describes the probability of driving to the different destinations from the origin. The obtained updated OD matrix was calibrated based on TFlowFuzzy approach. The manual counting data were used for the calibration of OD matrix. Finally the OD matrix was manually checked by the group of experts. The proposed methodology allows to estimate the OD matrix, which could be used for microscopic traffic modelling in case of dynamic routing application in the model.