Search results

Development of innovative methods and hardware gives additional opportunities for the solution of various tasks. For example, innovations allow increasing safety in places of crossing highways and railroads by means of detection methods of objects in the required control zone. The big risk of emergencies in places as a railway crossing is caused by a high speed and the inertial movement of the train. In this connection, there is no possibility of timely braking of the train and prevention of collision with an obstacle. At the present time, the biggest percent of accident rate belongs to uncontrollable railway crossings. On such railway crossings completely are absent the blocking means or the warning technical tools. Controlled railway crossings are equipped with the blocking means for the automobiles. Also, they are equipped with the systems of video supervision. But the current systems of video supervision have no function of informing the driver of the train in an emergency. Therefore, to priorities belongs development of technical requirements for a hardware and software system. The automatic analysis of a dynamic situation in borders of a railway crossing, transfer to the driver of the approaching train of the required information belongs to priorities too. This work describes briefly main algorithms of computer vision which allow distinguishing any object on a railway crossing. The example of algorithmic modeling of detection of contours and allocation of a background on video has been considered in this work. Received results led to the conclusion about a necessary adaptation of modern computer vision algorithms for the solution of a detection problem of various objects with the required accuracy and in various working conditions.

This article describes the approaches to extract from the video sequence of areas relating to movable objects, which may cause a dangerous situation at a railway crossing. The proposed approach will improve the quality of the mobile object allocation, which will reduce the time image processing to identify moving objects in order to determine the degree of risk. To assess the quality of algorithms was develop a software product and simulated motion capture.