Lane Detection System Based On Perspective Transform Algorithm

In this paper, we mainly discuss lane auto-detect system running on embedded system, there is a new digital image processing algorithm integrated in the system. Using the embedded system, either does it catch video stream, or does it figure out the lane on road accurately. We use digital peak value index algorithm and new perspective transform algorithm; both of them make this system have a good capability to find the lane on road in varieties of weather. This paper uses large proportion to describe the new perspective transform algorithm and how to realize this algorithm on embedded system. Because of using the new perspective transform and the knowledge of erasing noise on image processing, the system can figure out the lane line no matter how hard the environment on road is. In addition, the performance of this new perspective transform algorithm crosses over the other two algorithms which are about straight line detection algorithm and the edge of road detection algorithm. The software and hardware applying on this embed system use multi-thread technology, which makes the system have capability to deal with data in real time. Owe to an accurate digital camera and good algorithm applied on the system, driver will find the trend of lane which will extend, which will ensure the safe of driving on high wary, for example. Based on these algorithms and technologies introduced on the upper, lane auto-detect system can adapt different environments on road, that is either a important electric device to ensure safety drive, or a critical model which can be integrated in auto-drive system.