Single–camera vehicle speed measurement using the geometry of the imaging system

Abstract

In recent years, measuring the speed of vehicles by a single camera has been done using several methods including the road geometric information, the difference in the number of image pixels of moving objects, and the homographic mapping of the motion vector from the image plane to the global coordinate plane. In this paper, we present a new method based on the geometry of the imaging system and the definition of solid angle. Our method does not require 3D modeling; It has a lower computational cost than the existing methods; And it has an accuracy comparable to the other approaches. In this method, we use the video images taken with a single camera on the road to extract the license plate in the image. Then, using the geometric information of the system and the distance travelled by vehicles the speed is computed. The average relative error of the proposed method is 3.23% and the mean absolute error is 1.32 km/h, which is comparable to the available algorithms. Furthermore, the computational cost of our method is less than the existing ones, which makes it suitable for implementing on embedded systems.