Obstacle Detection In Front Of Automatic TrainsBy Linear Stereo Vision

Nowadays, the railways transport operators require anticipated obstacles detection on the tracks in order to avoid collisions. Such a system allows to : reduce the wasting of time of exploitation linked to the incidents; reduce travelling and waiting times for the passengers; improve passengers safety. In our system, we use optical sensors to detect the obstacles. Two matrix video cameras are installed in front of an automatic train in order to have two simultaneous views of the tracks. This configuration of artificial vision, called stereo vision, allows us to recover the three-dimensional scene in front of the train. The originality of the work carried out here is the use of linear procedures from the matrix video collected data. This way of working entitles us to define several surveillance planes which depend on the distance of detection required. First of all, we detect the rails in the acquired pictures, then we use this piece of information to define the Y-axis origin of the scene. All objects lying on the tracks on which Y co-ordinates are positive, will be considered as obstacles for the train. Transactions on the Built Environment vol 34, © 1998 WIT Press, www.witpress.com, ISSN 1743-3509 84 Computers in Railways