Investigating the reduction of the error caused by the counting of vehicle traffic in city license plate reader cameras by mathematical models (case study: Mashhad city)

Articles in Press

Document Type : Original Article

Authors

PhD Candidate in civil engineering, majoring in transportation planning, Faculty of Civil Engineering, Arts and Architecture, Tehran Science and Research

10.22034/road.2024.416435.2199

Abstract

One of the most important ITS tools are speed control cameras, and one of the most common cameras used in urban networks is the urban license plate reader (ANPR) cameras. Therefore, due to the importance of measuring the performance of urban license plate reader cameras in order to check and correctly understand the traffic situation of cities, A research has been done to identify existing errors and then provide a mathematical model to reduce camera errors.In this research, the sample data is related to traffic information recorded by two traffic cameras in Mashhad during the first week of June 2019.The errors identified in this research include: total camera error, license plate registration error, license plate letter registration error, city code registration error, violation registration error, whether the license plate is even or odd. Various factors have had a direct impact on the occurrence of these errors, including camera number, day of the week, time of the car, speed of the car, crossing line, density or volume of the car.In this research, in order to identify and reduce existing errors, two logistic regression and k-nearest neighbor classification models have been investigated and analyzed according to the existing parameters. In the total error, the most effective parameter is the speed and the nearest neighbor K model with 68% accuracy is the best classband for the total error of vehicle information registration

Keywords

Files

Share

How to cite

Statistics