Yeni bir sürücü uyarı sisteminin geliştirilmesi ve gerçek zamanlı olarak uygulanması

Most studies so far on Advanced Driver Warning Systems have assumed perfect knowledge of driving conditions, driver intents which require expensive sensor suite,data communication and processing. In this thesis, a fairly restrictive driver warning system supporting ?blind spot?, ?lane departure? detection and ?safe following distance? monitoring is studied. A vision-based active safety system is presented toprevent possible traffic accidents during lane change maneuvering and lane following tasks. The active system considers that the driver is always in-the-loop and in case oflane departure detection or lane change unintentionally into the path of the coming vehicle, visual and audio signal is generated to warn the driver.Proposed system brings a new approach to blind spot detection and lane recognition. Blob Detection and Hough Transform are used together for a robust detection of lanes. White pixel intensity method is proposed to be aware of the dangers in blind spot.Forward vehicles are easily detected by template matching algorithm.This new system does not only bring new approaches but also a lower cost basis comparing with the other after market products thanks to its three cameras and an onboard computer.