Railway and Public Guided Transit Properties often employ large numbers of video cameras to supervise critical areas and facilitate incident management. Capabilities of Central Control Staff is, however, limited to check the increasing number of CCTV images and so far automated image processing solutions had been insufficiently reliable. TelSys GmbH (a railway telematics company in Dresden, Germany) had therefore developed over the last seven years together with the University of Technology in Dresden and some public transport providers (subway of Berlin, subway of Prague) a robust solution to supervise automatically critical areas like tunnel entrances, station tracks or station platform edges. Also qualifications with German Railways and in Finland had been performed. The automatic image processing software reliably differentiates between trains (“permitted” objects) and objects that move from the platform into the tracks or move too close to otherwise prohibited areas. Object sizes, alarm times, reliability and safety requirements had been taken from the VDV 399 standard of the German Public Transport Operators Association. After years of reliability and safety research and demonstration the system is now in regular operation (stopping automatically incoming driverless trains if an object is detected in the track) and can be considered as the first safe video image processing system according to railway standards. Experiences, system architecture and principles as well as further development plans and planned demonstration installation in North America are discussed.