Method of determining vehicle speed according to video stream data

Abstract

This article presents a method of determining the speed of a dynamic object from the data of a video stream. The determination of speed occurs based on the analysis of repeated objects of the environment, the distance between which varies according to a previously known law, such as lamp posts, milestones or bump stops on roads, intermittent road marking lines, emergency exits in tunnels. The method of determining vehicle speed includes three steps: pre-processing of the frame, searching for a pattern in the image, calculating the average vehicle speed obtained on the basis of the number of repeated objects, taking into account the law of their repetition and the distance between them for some fixed time. The advantage of the proposed method is that there is no need to install any additional sensors on the dynamic object except video camera. The development of an additional method for calculating speed has a significant application value for the complexes of control of autonomous traffic, as the existing methods of determining speed have their disadvantages (not high enough accuracy, the requirement for the presence of additional equipment). This algorithm was studied on regularly recurring objects (lampposts) under different lighting conditions (day, night), the distance between repeated objects and different weather conditions. It was also compared with other methods of determining the speed of the data from the video stream, the results of which the proposed method has a significant advantage in accuracy as compared with analogs in the absence of interference with video.