Visual registration and tracking for traffic monitoring

This paper describes a work-in-progress system to process real-time information about conflictive traffic spots within a modern city. The system aims to integrate data about the status of particular street points through image sensors and several visual interpretation phases. This idea has been inspired by the increasing traffic problem in the Mexican city of Guadalajara and the recent advances in the field of telecommunications, networking and mobile phone technologies. The information is built over incoming data from traffic cameras whose images are processed by several visual computation algorithms in order to segment and track vehicles. The system calculates vehicle speed information that can be later observed through a virtual reality user interface. The visual algorithm includes visual extraction, motion detection and tracking modules which process color images that are captured through off-the-shelf cameras. The system represents one operational block of an integrated framework which can be later used by government agencies to overcome difficult events on the streets, such as accidents, abnormal people flow and such. The current status of the project and a quick review of its future contributions are presented at this paper.