Practical recommendation of crosswalk based on traffic surveillance

Traffic security is one of the most important problems to be improved for our daily environment, especially in Japan. Surveillance systems are effectively used to observe intrusion into dangerous and security area. They are well evaluated to monitor whether people enter such an area or not. A remote surveillance system has been developed, which supports users of mobile phones to facilitate network camera homing and electric-power appliance controlling. The surveillance system is designed based on a Linux PC with Web services, Open CV engine, remote monitoring camera, remote controlling sub-server and other facilities. Its image understanding facility by Open CV engine gives us information about intruders´ tracks at a monitored place. As the result, it is beneficial that we can exactly know moving direction of intruders. It is concluded that our surveillance system is efficient enough to observe a specific place, compute image understanding about it and mine some kind of useful information of intruder´s periodical behaviors. It can be applied to acquire data about intruders´ tracks when they go across relatively narrow streets. By means of such data, we can obtain information about suitable place to realize a necessary crosswalk which will comply with silent but real requests of intruders.