IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. X, NO. X, MONTH 2016 1 A Video-Based System for Vehicle Speed Measurement in Urban Roadways Diogo C. Luvizon, Bogdan T. Nassu and Rodrigo Minetto Abstract—In this paper, we propose a non-intrusive, video- based system for vehicle speed measurement in urban roadways. Our system uses an optimized motion detector and a novel text detector to efficiently locate vehicle license plates in image regions containing motion. Distinctive features are then selected on the license plate regions, tracked across multiple frames, and rectified for perspective distortion. Vehicle speed is measured by comparing the trajectories of the tracked features to known real world measures. The proposed system was tested on a data set containing approximately five hours of videos recorded in different weather conditions by a single low-cost camera, with associated ground truth speeds obtained by an inductive loop detector. Our data set is freely available for research purposes. The measured speeds have an average error of -0.5 km/h, staying inside the [-3,+2] km/h limit determined by regulatory authorities in several countries in over 96.0% of the cases. To the authors’ knowledge, there are no other video-based systems able to achieve results comparable to those produced by an inductive loop detector. We also show that our license plate detector outperforms other two published state-of-the-art text detectors, as well as a well-known license plate detector, achieving a precision of 0.93 and a recall of 0.87. Index Terms—vehicle speed measurement; license plate detec- tion; feature tracking; vehicle motion detection. I. I NTRODUCTION Systems for vehicle detection and speed measurement play an important role in enforcing speed limits. They also provide relevant data for traffic control. Those systems are divided in intrusive and non-intrusive [1]. Intrusive sensors, usually based on inductive loop detectors, are widely used, but have complex installation and maintenance, accelerate asphalt deterioration, and can be damaged by wear and tear. Non-intrusive sensors, which include laser meters and Doppler radars, avoid these problems, but are usually more expensive and require frequent maintenance. As digital cameras become cheaper and able to produce images with higher quality, video-based systems can become a lower cost alternative for non-intrusive speed measurement. In fact, existing systems are often connected to video cameras [2] that record the license plates of vehicles that exceed the speed limit — thus, the infrastructure for such systems is already available in most cases. In this paper we describe the pipeline for a non-intrusive video-based system for vehicle speed measurement in urban roadways. Our goal is measuring vehicle speeds with accuracy Diogo C. Luvizon, Bogdan T. Nassu and Rodrigo Minetto are with the Department of Informatics, Federal University of Technology - Paran´ a (UTFPR), Curitiba, Brazil (e-mail: diogoluvizon@gmail.com, nassu@dainf.ct.utfpr.edu.br, rminetto@dainf.ct.utfpr.edu.br) Manuscript received X; revised X. comparable to that obtained by a system based on inductive loop detectors. The input video is captured by a single fixed overhead camera, positioned so that the rear license plate of vehicles in three adjacent lanes are clearly visible, as shown in Fig. 1. A sample image from this setup is shown in Fig. 2. This setup allows the same images to be used for both speed measurement and license plate identification (e.g. for locating stolen vehicles, or in the case of a speed limit violation). Video camera Lane 3 Lane 2 Lane 1 Ground truth aquisition by inductive loop detectors 5.5 m Fig. 1. System setup. Fig. 2. Sample image captured by our system. We make some assumptions about the scene and the prob- lem domain: video frames are equally spaced in time; each lane lies on a plane; the vehicles move at a constant speed and with a straight trajectory from the lower to the upper part of the image; and the license plates are at approximately the same distance from the ground. These assumptions allow us to measure vehicle speeds without modeling the 3-D space, or requiring precise camera calibration or positioning. The proposed system works by tracking sets of distinctive features extracted from image regions around each vehicle’s